c-Fos expression following context conditioning and deep brain stimulation in the bed nucleus of the stria terminalis in rats

Deep brain stimulation (DBS) in the bed nucleus of the stria terminalis (BST), a region implicated in the expression of anxiety, shows promise in psychiatric patients, but its effects throughout the limbic system are largely unknown. In male Wistar rats, we first evaluated the neural signature of contextual fear (N = 16) and next, of the anxiolytic effects of high-frequency electrical stimulation in the BST (N = 31), by means of c-Fos protein expression. In non-operated animals, we found that the left medial anterior BST displayed increased c-Fos expression in anxious (i.e., context-conditioned) versus control subjects. Moreover, control rats showed asymmetric expression in the basolateral amygdala (BLA) (i.e., higher intensities in the right hemisphere), which was absent in anxious animals. The predominant finding in rats receiving bilateral BST stimulation was a striking increase in c-Fos expression throughout much of the left hemisphere, which was not confined to the predefined regions of interest. To conclude, we found evidence for lateralized c-Fos expression during the expression of contextual fear and anxiolytic high-frequency electrical stimulation of the BST, particularly in the medial anterior BST and BLA. In addition, we observed an extensive and unexpected left-sided c-Fos spread following bilateral stimulation in the BST.

collected automatically from the Startle Reflex software. Based on their Pre-test startle values, rats were matched into two groups with comparable startle responses (ANX, n = 9; CTRL, n = 7).
Training. On day 3, after the 5-min acclimation phase, rats received 10 electrical foot shocks (0.8 mA, 250 ms; or 0 mA in CTRL animals) with a variable ITI of 60 -180 s. At this stage, the rats were conditioned to the context. The total duration of this session was 30 min.
Post-test. On day 4, the animals were tested using a 20-min protocol identical to that of the Pre-test. During Post-test, ANX animals were expected to express anxiety in the context where they previously received electrical shocks, as quantified by increased freezing during acclimation and startle potentiation.

Experiment 2
The standard conditioning procedure was followed. Following the Pre-test, animals were assigned to two groups, STIM (n = 17) and SHAM (n = 14). All rats received shocks during Training and were therefore conditioned to the context. STIM animals received electrical BST stimulation prior to and during the Post-test, as detailed below.

Electrical stimulation
For STIM animals in Experiment 2, high-frequency electrical stimulation of the bilateral BST was initiated in a home cage 1 h before and continued during the Post-test. We used a fixed frequency of 130 Hz and pulse width of 220 µs, whereas the amplitude was determined for every animal individually. Amplitudes were increased gradually until persistent side effects occurred (such as excessive shaking, urination, muscle spasms). Next, amplitudes were set just below this threshold. Stimulation of the bilateral BST was performed using two separate, isolated channels and identical parameters. We used monopolar stimulation (the reference was one of the fixation screws on the skull) with biphasic, counterbalanced pulses to avoid tissue damage.

Immunohistochemistry
In Experiment 1, six representative rats of both groups were selected to reflect the group average for freezing and startle measurements as accurately as possible (ANX: n = 6, CTRL: n = 6). This sample size is in accordance with what is commonly used in the c-Fos literature (typically n = 4-6 per group). Note that brain tissue sectioning and subsequent c-Fos analysis was only conducted for this subset. For Experiment 2, to increase power, all animals with correct electrode placement were included (STIM: n = 14, SHAM: n = 11).
Tissue preparation. Two hours after initiation of the Post-test, animals were deeply anesthetized by an intraperitoneal injection of pentobarbital (2.5 ml, Nembutal, CEVA Santé Animale, Brussels, Belgium). Next, the animals were perfused with saline and a 4% paraformaldehyde dilution in phosphate-buffered saline (PBS). Brains were removed and post-fixed for 24 h in 4% paraformaldehyde. Subsequently, samples were rinsed in water during 24 h and stored at 4°C in PBS. Fifty-µm thick free-floating serial sections were prepared on a Vibratome (Microm HM 650 V, Thermo Scientific, Walldorf, Germany) and collected in 24-well plates.
Image acquisition. Projection images of the regions of interest (ROIs) were acquired through an inverted FV1000 confocal microscope (IX81, Olympus; Aartselaar, Belgium) using a 20X objective (NA: 0.75) at a resolution of 512x512 µm. Upper and lower limits of the slice were determined in the Z-plane. Pictures were acquired as three subsequent z-stacks at 3-µm intervals covering the middle section of the slice, resulting in 9-µm reconstructions. For each structure of interest, three pictures were taken for each hemisphere. Image acquisition was done with FluoView10 software (Olympus, Aartselaar, Belgium) and comprised c-Fos, NeuN and GFAP channels at fixed intensity settings. For all c-Fos imaging, identical laser settings were applied. A low laser intensity of 5% was used, thereby minimizing any risk of bleaching during scanning, and an offset of 5% was used to filter out background fluorescence.
Image analysis. Images were imported in ImageJ (Rasband, ImageJ, National Institutes of Health, Bethesda, MD, http://imagej.nih.gov/ij/, 1997-2014) and ROIs were delineated as 200x200 µm squares situated in each structure of interest ( Fig. 1 in main article). For each ROI, a neuronal cell mask was obtained by auto-thresholding the NeuN image and detecting cells with circularity levels exceeding 20%. These cells were then automatically delineated and transferred to the c-Fos image, where intensity levels were detected and corrected for total neuronal cell size. Intensity levels were expressed on an arbitrary scale from 0 to 4095. For each unilateral structure, intensity levels were averaged over three consecutive slices. Note that we did not use the number of c-Fos positive neurons as a measure of activity, since most neurons expressed c-Fos under control conditions in our first experiment (i.e., in non-conditioned animals). Instead, we measured intensity levels, which provide information on the degree of activity rather than the number of active cells.

Supplementary c-Fos data
In the main article, we described two experiments in which we evaluated neuronal activity in the limbic system through c-Fos expression on the protein level. In Experiment 1, we measured c-Fos intensities in regions of the anxiety network in animals without implanted electrodes: context-conditioned rats (ANX) and control animals that did not receive any shocks (CTRL). In Experiment 2, both groups were context-conditioned and, during the post-test, received high-frequency electrical stimulation of the BST (STIM) or no stimulation (SHAM). All findings were discussed in the main Results section, but only a selection of graphs was shown for clarity. Here, all remaining bar plots and tables are provided for the sake of completeness.  Table 1. Significances of betweengroup differences are listed in the right-hand column. Withinand between-group differences were only observed in the BLA and STMA, respectively. The shell of the NAc showed a significant effect of 'Hemisphere' (F(1,10) = 6.77; p = 0.03), with the left side expressing more c-Fos compared to its contralateral counterpart, in both groups. Again, this finding underlines the importance of evaluating neural activity per hemisphere, rather than analyzing only one hemisphere, or pooling information from both sides of the brain.

Supplementary Figure 1: c-Fos intensities in Experiment 1. Neuronal c-Fos expression is shown for the STL (top panel), NAc core (middle panel) and NAc shell (bottom panel).
Data are shown as individual data points and means per hemisphere, for CTRL (n = 6) and ANX (n = 6) animals, *p < .05. STL: lateral division of the bed nucleus of the stria terminalis, NAc: nucleus accumbens.

Experiment 2
Bar plots with individual data points and means for all ROIs (basolateral amygdala, nucleus accumbens core and shell, infralimbic and prelimbic cortex) are shown in Suppl. Fig. 2. As mentioned in the main Results, we observed a pronounced increase in c-Fos intensities in all left-sided ROIs in animals receiving bilateral electrical stimulation of the BST. In addition, the BLA showed a strong increase in c-Fos expression in the right hemisphere in animals receiving BST stimulation versus sham stimulation (Sidak post-hoc test, t(46) = 3.19, p = .005).

Additional analyses
To gain more insight into the remarkable increase of c-Fos intensity in the left hemisphere accompanying bilateral BST stimulation in Experiment 2, we conducted some exploratory analyses, going beyond the predefined ROIs.
Hemisphere-wide evaluation of c-Fos spread To assess hemisphere-wide c-Fos expression, we imaged full coronal sections (+2.40, +1.20, 0.00, -2.40 mm with respect to bregma) using a fluorescent microscope with 5X magnification (Zeiss Imager Z1) ( Fig. 6 in main article). Note that these images were not acquired in a confocal manner and c-Fos expression therefore represents the total thickness of the slice (50 µm). In additional contrast to all other analyses, we did not use a neuronal cell mask to quantify c-Fos, given that a neuronal cell count was not possible for this hemisphere-wide analysis, because of extremely high numbers compared to a 200x200 µm ROI analysis. For exploratory quantification of the clearly visible left-right difference, each hemisphere was delineated manually as a ROI in ImageJ, average c-Fos intensities corrected for area size were calculated for each hemisphere, and these were plotted as the ratio of left and right intensities (Suppl. Fig. 3). We found that STIM animals had higher lateralization ratios compared to SHAM animals in every coronal section that we investigated. Nominal values were highest in both sections anterior to bregma. None of the 95% confidence intervals of STIM animals comprised the 100 % mark, suggesting lateralization, which was, however, less pronounced in the section posterior to bregma. Finally, SHAM animals showed no hemisphere-wide lateralization and the confidence interval comprised the 100 % mark for every section.

Supplementary Figure 2: c-Fos intensities in Experiment 2. Neuronal c-Fos
Quantification of c-Fos levels adjacent to electrode tips Next, we investigated in more detail the c-Fos expression in the immediate vicinity of the electrode tips, to verify that left and right BST received comparable electrical stimulation. We expected similar levels of c-Fos in both hemispheres as a result of direct (i.e. non-trans-synaptic) effects of the electrical current. We sectioned the BST of all animals and analyzed cross-sections through (or as close as possible to) the bilateral electrodes (Suppl. Fig. 4). Neuronal c-Fos expression in ten 50µm bins surrounding the electrode was quantified using a custom-made Image J plugin. As described in the main article, there was a clear increase in c-Fos expression around the electrode tip in STIM versus SHAM animals, without significant differences between hemispheres (Fig. 7 in main article).
A more detailed analysis, per 50 µm, did indicate some hemispheric differences in STIM animals, but if anything, c-Fos intensity was slightly higher in the right than in the left hemisphere, close to the electrode tip (Suppl. Fig. 5). More specifically, the two-way RM-ANOVA in the STIM group indicated no effect of 'Hemisphere' nor 'Distance', but a significant interaction between both (F(9,117) = 3.89, p < .001), with the difference situated within the 150 µm closest to the electrode (Sidak posthoc tests on 50-µm bins, p < .01). The ANOVA in the SHAM group showed no main effects nor interaction. Note that this fine-grained analysis per 50 µm should be interpreted with sufficient caution because manual delineation of the electrode tip was not always straightforward, as mentioned in the main article. Nonetheless, these exploratory analyses do suggest that STIM rats did not receive less electrical stimulation in the right than in the left BST, in line with the bilateral stimulation protocol that we applied.

Supplementary Figure 3: Lateralization ratio of hemisphere-wide c-Fos intensities, corrected for area size.
Values above 100% indicate higher c-Fos intensities in the left hemisphere than in the right hemisphere. Data are shown as means and 95% confidence intervals, for both SHAM (n = 11) and STIM (n = 14) animals, at different positions (expressed in mm) relative to the coronal bregma slice (= 0.00 mm).