Major depressive disorder (MDD) will affect one out of every five people in their lifetime. It is also the leading cause of disabilities worldwide. Unfortunately current treatments are ineffective in a large subset of patients and pathogenesis of MDD has yet to be completely understood. Alterations within the peripheral immune system and overactivation of proinflammatory cytokines have been associated with mood disorders for decades. Monocytes may even traffic to the brain via the bloodstream suggesting that the blood-brain barrier (BBB) is compromised. Evidences from clinical studies such as an altered ratio of cerebrospinal fluid to serum of various markers in depressed patients also suggest BBB dysfunction. In this pilot study we compared BBB integrity and brain connectivity in vivo in mice following chronic social defeat stress [1], an established rodent model of MDD.3 groups of 5 C57Bl/6 mice were scanned in this pilot study. 10 mice were subjected to a 10-day social defeat stress paradigm (stressed mice). Stressed mice were exposed to physical interaction with a novel larger CD-1 aggressor every day for 10 minutes followed by sensory interaction for 24 hours. Mice were then separated into 2 groups based on their social interaction phenotype which was conducted 24 hours after the last defeat. 5 control mice were housed in the same room but not subjected to the stress paradigm (unstressed control mice). All mice were scanned on a Bruker Biospec 7T/30 scanner using a 4 channel phased array coil 2-4 days following the social interaction test. The following protocols were obtained: T1 weighted 2D-FLASH, PGSE-DTI (b-value=1200s/mm2, FOV=15mm, MTX=128x128, slice-thickness=0.5mm,22 slices, 30 directions, 5 b=0), resting state fMRI (GE-EPI: FOV=20mm, Matrix=64x64, slice thickness=0.5mm, 20 slices, TE=20, TR=1s, Imaging time=10 minutes (600 volumes), Manual shimming was performed for every fMRI scan. T1 and T2 weighted images were first acquired and the mouse was taken out and contrast agent (Gd-DTPA) was administered through the intraocular pathway and the mouse was then rescanned. Resting state and DTI scans were obtained on a separate date to minimize the effects of lingering contrast agents. Anesthesia used for fMRI was Dexmedetomidine infused through a subcutaneous catheter and infusion pump. All other scans were performed using isoflurane.Regions of interests were defined over the striatum, thalamus, hippocampus, nucleus acumbens, and various portions of the sensory/motor cortices using in-house developed matlab based software. Statistical differences were computed using Statistica (Statsoft Inc.). Resting state fmri scans were coregistered to T2 anatomical scans. The brain was identified on the T2 scans and a binary mask was generated using FSL software (www.fmrib.ox.ac.uk/fsl). This mask was then used to brain extract the fMRI data. A study specific anatomical template was generated from the 15 mice. All fMRI scans were then transformed and coregistered to this template. Independent component analysis was performed using MELODIC (FSL) with 20 networks. Dual regression analysis was performed to compare the different groups. DTI data was processed using FSL. Fractional anisotropy (FA) maps were computed and region of interest analysis was performed on the following structures: corpus callosum, cingulum bundle, external capsule, internal capsule, fimbria and the tractus retroflexus. Statistical group comparisons and correlations were computed using statistica.Several regions in the brain showed significant signal differences when the contrast agent was administered. This was detected in the hippocampus, sensory/motor cortices and thalamus. The same regions were also significantly correlated with the Social Interaction (SI) index. Resting state fMRI showed significant increased activity in the default mode network of the susceptible mice when compared with resilient or control. No differences were detected between controls and resilient mice. Fractional anisotropy ROIs did not produce any significant group differences, but when all ROIs were pooled together to represent whole brain FA index, the resilient group has significantly high FA values.In this pilot study we have performed a battery of imaging protocols on a social defeat model in rodents. We have shown that several regions may have compromised BBB in the susceptible phenotype. In addition the anterior cingulate in the default mode network showed hyperactivity in the susceptible mice. This result is consistent with hyperactivity in PTSD patients as reported in fMRI studies [2]. The higher whole brain fractional anisotropy in the resilient group might signify better structure connectivity that predisposes them to be resilient. Further studies with increased number of animals will need to be conducted to confirm these results.