Converging Prefronto-Insula-Amygdala Pathways in Negative Emotion Regulation in Marmoset Monkeys

Background Impaired regulation of emotional responses to potential threat is a core feature of affective disorders. However, while the subcortical circuitry responsible for processing and expression of fear has been well characterized, the top-down control of this circuitry is less well understood. Our recent studies demonstrated that heightened emotionality, as measured both physiologically and behaviorally, during conditioned fear and innate/social threat was induced, independently, by excitotoxic lesions of either the anterior orbitofrontal cortex (antOFC) or ventrolateral prefrontal cortex (vlPFC). An important outstanding question is whether the antOFC and vlPFC act on common or distinct downstream targets to regulate negative emotion. Methods The question was addressed by combining localized excitotoxic lesions in the PFC of a nonhuman primate and functional neuroimaging ([18F]fluorodeoxyglucose positron emission tomography) with a fear-regulating extinction paradigm. Marmoset monkeys with unilateral lesions of either the antOFC or vlPFC were scanned immediately following exposure to a fearful or safe context, and differences in [18F]fluorodeoxyglucose uptake were evaluated. Results [18F]fluorodeoxyglucose uptake in the insula and amygdala of the intact hemisphere was significantly increased in response to the fearful context compared with the safe context. Such discrimination between the two contexts was not reflected in the activity of the insula-amygdala of the antOFC or vlPFC-lesioned hemisphere. Instead, uptake was at an intermediate level in both contexts. Conclusions These findings demonstrate that the distinct control functions of the antOFC and vlPFC converge on the same downstream targets to promote emotion regulation, taking us closer to a mechanistic understanding of different forms of anxiety.

nuts on the weekends. Water was available ad libitum.

Telemetry Recording System
To measure blood pressure (BP) changes remotely in freely moving animals, a PhysioTel Telemetry System (Data Sciences, Inc. (DSI), St Paul, Minnesota, USA) was used. The system consisted of five basic components: 1) An implantable transmitter (TA11PA-C40, DSI) which continuously detected and transmitted BP from within the animal via radio-frequency signals; 2) A receiver (RPC-1, DSI) located underneath the behavioral testing box, which received the digitized information from the implanted transmitter and relayed the data for subsequent translation; 3) A calibrated pressure output adapter (R11CPA, DSI) with an ambient pressure reference monitor (APR-1, DSI) to convert the absolute pressure measured by the through a stainless-steel cannula (30 gauge). The cannula remained in place for 4min, after which it was slowly withdrawn from the brain. The skin was sutured and covered with a protective barrier (Germoline New Skin; Bayer, Newbury, UK), and dexamethasone phosphate (0.2ml i.m.; Fauling Pharmaceuticals plc, Warwicks, UK) was administered to avoid the unlikely event of tissue inflammation. Non-steroidal analgesics (0.1ml Metacam oral; St. Joseph, MO, USA) were given for 3 days after surgery at 24-h intervals. The animals had at least a 2-week recovery period.

Port Implant Surgery
Immediately prior to the surgery, the animal was intubated and anesthetized, following the same procedure described above. The animal was placed on a surgical table in a prone position. A small incision was made below the shoulder blades at right angle to the axis of the spine on the animal's back where a soloport would be placed. Another small incision was made on the neck to expose the jugular vein. A catheter attached to the port was then threaded under the skin from the back towards the neck. The port was placed in the skin pocket on the back. Through a small cut made in the jugular vein, the open end of the catheter was inserted in the direction of the heart. The catheter was glued to the vein with Vetbond (M3 Animal Care Products, MN, USA) and the incisions on the back and neck sutured. Following the surgery, analgesics (0.1ml Metacam oral; St. Joseph, MO, USA) and antibiotics (0.25ml Synulox oral; Pfizer Ltd., Kent, UK) were given daily for 3 and 7 days respectively. The soloport was flushed with Hepsal post-surgery on days 1, 3, 6, 10, 15 and then weekly. The animals had at least 10 days recovery period before behavioral testing began.

Cardiovascular Implant Surgery
One day prior to surgery, animals received prophylactic antibiotic treatment: 0.25ml Flagyl-S (40mg/ml metronidazole; Winthrop Pharmaceuticals., Guildford, UK) and 0.25ml Synulox (50mg/ml clavlanate-potentiated amoxicillin; Pfizer Ltd., Kent, UK). For implantation of telemetry probes, marmosets were premedicated with ketamine hydrochloride (sedative, 0.1ml of a 100 mg/ml solution, intramuscular (i.m.); Amersham Pharmacia and Upjohn, Piscataway, NJ, USA) and carprofen (prophylactic analgesic, 0.03ml, subcutaneous (s.c.)), and anesthetized by isoflurane intubation (flow rate 2-2.5%; IsoFlo, Abbott Laboratories, Abbott Park, IL, USA). The animal was placed in a supine position onto a sterile drape and the limbs were secured with masking tape to allow unrestricted access to the abdomen. Under aseptic conditions, a 4-6cm midline abdominal incision was made using scissors to allow a clear view of the aorta from the upper portion of the vessel down to the bifurcation of the aorta to the renal arteries. The aorta was then carefully dissected from the surrounding fat and connective tissue. Once isolated, the lower portion of the aorta, just above the bifurcation, was lifted and a cotton thread, approximately 8cm in length, was passed underneath. The two ends were then clamped together with forceps to 1) lift the vessel for implantation and 2) to exert a small amount of tension to prevent blood reflux after blood flow to the area was restricted during catheterisation. Once the vessel was clear, to restrict blood flow, a finger was used to apply pressure to the upper most portion of the aorta and slight tension was placed on the thread at the base. Using a 23-gauge needle (bent at 60°, bevelled edge upwards), the vessel was punctured just above the bifurcation and the tip of the catheter was inserted using a catheter introducer. The catheter was then passed up the length of the vessel until approximately 30-40mm of the tubing was contained within the vessel. Once correctly positioned, the area was thoroughly dried and Vetbond (M3 Animal Care Products, Minnesota, USA) tissue adhesive was applied to the puncture site. After integrity of the seal was established, a cellulose patch was placed over the entry site and fixed in position with additional adhesive. Following implantation, the thread and retractors were removed and the abdominal cavity was moistened with sterile saline. The device body was then secured in position by incorporating the tabs on the implant into the muscle wall by using non-absorbable sutures (Ethilon 3-0 W; Ethicon Inc., Georgia, USA). After the closure of the muscle wall, the skin was closed using absorbable sutures (3-0 Vicryl W9444; Ethicon Inc., Georgia, USA) and Vetbond was applied to each stitch to ensure that the abdomen was completely sealed.
Antibiotics, 0.25ml each of Flagyl and Synulox, were also administered orally for 10 days post-surgery to protect against intestinal infection. Marmosets had a two-week recovery period before testing began.

MR and PET Imaging
MRI -An in-house custom-built quadrature birdcage coil was used for signal transmission and reception. Images were acquired with a matrix of 256 × 200 over a field of view of 6.40cm × 5.00cm yielding in plane resolution 250μm with 125 slices of 250μm. The repetition time was 11.75s with effective echo time of 23.5ms. Three repetitions were acquired at a bandwidth of 34.7kHz and averaged for a total scan time of 21min 44s.
PET scan -On the day of scan, the animals received no breakfast in order to lower blood glucose concentration and hence increase the transport of FDG into brain tissue, thereby increasing the cerebral FDG signal and hence reducing statistical noise in the PET image.

Animals were placed in a test box approximately 3 minutes after a bolus injection of 71±14
MBq of FDG subcutaneously through the solo port. After 30 min of the behavioral paradigm described below, the animal was immediately intubated and anaesthetized following the procedure described above. The animal was then placed on the heatpad on the scanner bed and attached to monitoring equipment. Heart rate, SpO 2 and respirations were monitored continuously. The bed of the scanner was then positioned to locate the brain in the center of the PET scanner field of view, where both sensitivity and resolution are optimal. For consistency, PET data acquisition started 70min after the FDG injection and lasted for 45min.
The energy and coincidence timing windows used were 350-650 keV and 6 nsecs, respectively.
The list mode data were histogrammed into 9 × 5min 4D sinograms, and then reconstructed using Fourier re-binning (FORE; (2)) followed by the 2D ordered subsets expectation maximization (OSEM; (3)) algorithm installed on the scanner (6 iterations, 16 subsets). As post-injection transmission scanning was not feasible, attenuation correction used a mean non-attenuation corrected FDG image to determine a body outline, within which a uniform attenuation coefficient (0.096 cm -1 ) was ascribed. This was combined with a standard attenuation map of the carbon fiber bed determined from transmission scanning. The combined attenuation map was forward projected using software installed on the scanner to produce an attenuation correction factor sinogram, and image reconstruction was repeated with attenuation correction applied. Corrections were also applied for randoms, dead time, normalization, sensitivity, and decay.

Processing of PET Data
Using SPM8 (Wellcome Trust Institute for Neurology, UCL, UK), the MR image of each subject was registered rigidly to a colony-specific structural template produced during previous studies (4). This provided consistent alignment of the MR scans for intra-subject manual rigid registration of PET to MR, which for each PET scan used the mean PET image across all frames. The realigned MR was non-rigidly registered (affine and non-linear) to the structural template using ANTS (5), and this transformation was also applied to the mean PET image For voxel-wise analysis, SUVRc from animals receiving a lesion in the left hemisphere were flipped about the midline such that for all images the intact side of the brain appears on the left. To mitigate against residual registration error and increase data normality, each SUVRc image was smoothed using a Gaussian kernel of 1mm 3 . The kernel was locally adapted to include only those voxels within a brain mask to avoid contamination from extracerebral tissue signal.

Behavioral Analysis
In contrast to the wide array of behavioral responses displayed by a marmoset to a rubber snake when tested in their home cage environment (7), the behavioral repertoire observed in the relatively confined space of the carrying box in the test apparatus was limited. In particular, the animals stayed relatively immobile during the fear-inducing condition and sometimes also during the safety condition. Although not consistent across animals, rearing was seen periodically in the presence of the snake stimulus but no vocalizations were made. The only behavior consistently seen in all animals was that they remained as far away as possible from the location of the snake. Therefore, the floor of the carrying box was divided into a near half and a far half and the duration of time spent in each location was compared in the fear and safety conditions. It appeared that in the safety condition, in the absence of the snake, marmosets tended to prefer to sit in the 'near' position rather than distribute their time evenly across 'near' and 'far' conditions. Speculatively, the near position was close to the light, open space within the chamber (see Figure 1 in main article), which, without the presence of the snake, may have made it a more attractive place in which to sit.

Cardiovascular Analysis
Blood pressure (BP) data was transmitted by an implanted telemetry probe to a receiver for analysis using Spike2 (Version 7.01, CED) as described previously (8). Outliers and recording failures were removed (values above 200mmHg, below 0mmHg or other abnormal spikes).
Data collection was reliable overall, however gaps of less than 0.4s were replaced by cubic spline interpolation and gaps of more than 0.4s were treated as missing values. Systolic BP and diastolic BP events were extracted as local maxima and minima for each cardiac cycle.
Systolic BP at individual time-points were binned into 1s intervals and then calculated as an average over each 1s bin.

Histological Preparation
All marmosets were euthanized with Dolethal (1 ml of a 200 mg/ml solution, pentobarbital sodium, i.p.; Merial Animal Health, Essex, U.K.). Animals were then perfused transcardially with 500 ml of 0.1 M PBS (pH 7.4), followed by 500 ml of 0.4% formaldehyde-buffered solution, washed through over 10 min. The entire brain was removed and placed in fixative solution overnight before being transferred to a 30% sucrose solution in 0.01 M PBS for a minimum of 48h. The brain was then sectioned using microtome into 60µm thick slices. Each brain section was mounted on a slide and stained with cresyl blue.

Statistical Analysis
For all parametric analyses, i.e. factorial ANOVA, Kolmogorov-Smirnov test was used to test the normality assumption, and Levene's test was used to examine the homogeneity of variance. The assumptions were satisfied unless otherwise noted.

Control Group
When the behavioral pattern in response to the fear or safety conditions was compared between the two lesion groups (antOFC n=7, vlPFC n=4) and the telemetry control group (n=3), no significant group difference was found [three-way ANOVA: Condition (Fear vs Safety) x Distance (Near vs Far) x Group (antOFC vs vlPFC vs Telemetry control): No significant interaction Condition x Distance x Group F(2,11)=1.658, p=0.235].

Right-Unilaterally Lesioned Animals
The FDG uptake in response to the fear-inducing condition (averaged across Fear 1 and 2) proportional to the safety condition were compared between the animals that had the unilateral lesion in the left or right hemisphere. For the antOFC group (Left=4, Right=3), no significant difference was found either in the 'p<0.001 insula-amygdala cluster' [one-way ANOVA (Left vs Right): F<1; Amygdala cluster F<1; Insula cluster F<1] (Supplemental Figure S1-A-i). The vlPFC group had three right-hemisphere lesioned animals but only one left-hemisphere lesioned animal. Calculation of the 95% confidence limits for the group of three righthemisphere lesioned animals showed that the left-hemisphere lesioned animal fell within the range in the 'p<0.001 insula-amygdala cluster' [Left animal: 5.16%; Right animals: 11.61% ± 7.67%; Amygdala cluster L: 6.82%; R: 12.26% ± 11.29%; Insula cluster L: 4.97%; R: 11.58% ± 7.95%], suggesting that the left-hemisphere lesioned animal was not differentiable from the right-hemisphere lesioned animals (Supplemental Figure S1-B-i).

Animals
Similarly, the behavioral response (distance from the snake) between the left-and rightlesioned animals for the antOFC group showed no effect of hemisphere [three-way factorial For the vlPFC group, the left-hemisphere lesioned animal fell within the range of 95% Confidence Interval for the group of three right-hemisphere lesioned animals during the fearinducing condition [Near: L: 24.25%; R: 33.57% ± 41.48%] (Supplemental Figure S1-B-ii).
Note: The 95% confidence limits for the 3 right-lesioned animals in the vlPFC group during 'snake presentation' are large because one of the 3 animals had a tendency to move about during the dark period and freeze as soon as the light came back on and the snake could be seen. In three out of the four snake presentations this animal stopped moving on the near-far border but just within the near sector so their 'far' scores are much lower than the other two animals.

Fear Exposures
There were no significant differences in the behavioural response (distance from the snake) between the two conditions (Supplemental Figure S2 i: antOFC group, ii: vlPFC group).
Comparison of the FDG uptake of the significant insula-amygdala cluster (p<0.001) in the intact hemisphere for the first exposure to the fear-inducing stimuli (Fear 1) and the replication session (Fear 2) revealed no significant differences (Supplemental Figure S3, left upper figure).

No Difference in FDG Uptake Or Behaviour Was Found Between Males and Females
Male and females were compared for the percentage change in the FDG uptake values from the safety to the fear condition in the intact side. No difference was found in any of the clusters: To test the robustness of the findings we performed a bootstrap procedure. For each animal, mean values of SUVR were extracted from the insula-amygdala (p<0.001) cluster on the intact and lesioned sides. The mean values for the safety condition were subtracted from the mean fear scan values. Four separate bootstrap procedures were followed, each with 1,000 samples.
For the intact hemisphere, resampling was performed from all animals. On the lesioned side, samples were repeated from a) all animals, b) only antOFC-lesioned animals, c) only vlPFClesioned animals. Percentile confidence intervals of 95% were calculated from each sample.
As can be seen in Supplemental Figure S4, the intact hemisphere of all animals showed differential activation in fear compared to safety conditions but there was no such differential activation on the lesioned side, irrespective of lesion site. Furthermore, there was no evidence that the response to fear vs. safety on the lesioned side varied depending on whether the lesion targeted antOFC or vlPFC. Figure S1. Comparison of the FDG uptake in response to the fear-inducing condition (averaged across Fear 1 and 2) proportional to the safety condition of the animals that had the unilateral lesion in the left or right hemisphere for the antOFC group (A-i) (Left=4, Right=3) and vlPFC group (B-i) (Left=1, Right=3). Comparison of the behavioral response (distance from the snake) between the left-and right-lesioned animals for the antOFC group (A-ii) and vlPFC group (B-ii). No significant difference was found between the left and right lesioned animals in any of the measures. Figure S2. Comparison of the behavioural response (distance from the snake) during the first exposure to the fear-inducing stimuli (Fear 1; grey bars) and the replication session (Fear 2, black bars) for the antOFC group (i) and the vlPFC group (ii). There was no statistically significant difference between the first exposure and replication session. Error bars indicate standard errors. Figure S3. Adjusted means of the FDG uptake scores in the intact (left-hand side) and lesioned hemispheres (right-hand side) for the significant insula-amygdala cluster [p<0.001] (upper figures), the same cluster masked by the amygdala ROI (middle figures) and the remaining insula cluster (lower figures), across the three conditions (first fear-induction; Fear 1, safety, fear replication; Fear 2). Adjustment was performed by fitting a general linear model with effects of Subject and Condition. Data are shown with individual subject effects removed and replaced with a mean subject effect. Error bars show 95% confidence intervals. Figure S4. Percentile 95% confidence intervals. In the contralateral intact hemisphere the difference in FDG uptake during fear compared to safety was significantly different from '0'. In contrast, all ipsilateral lesioned hemispheres (All), antOFC lesioned only (OFC) or vlPFC lesioned only (vlPFC) did not differ from '0' and were significantly lower than the intact hemisphere.