Schizophrenia is a severe mental disorder that afflicts 1% of the world’s population. However, the cause of this disorder remains unknown. Increasing evidence points toward a neurodevelopmental mechanism for schizophrenia, which implicates genes involved in neuronal proliferation, migration, or synapse formation. These developmental processes are highly regulated by polysialylated (PSA) neural cell adhesion molecule (NCAM). The mouse model of PSA-NCAM deficiency (PSA-KO) presents a schizophrenic like phenotype¹, as well as a decrease anxiety and sociability².

In this pilot study, we investigate microstructural changes by ex-vivo diffusion MRI (dMRI) in PSA-KO mice brain. In combination to conventional diffusion tensor imaging (DTI), biophysical diffusion compartment model was used to get a better insight on the cellular organization. The Neurite Orientation Dispersion and Density Imaging model (NODDI) was chosen³, due to its easily translation to clinical scanner and its applications in developmental study⁴.

Animals: Mice were sacrificed at 6 months old after behavioral testing. Brains were fixed in PFA by intra-cardiac infusion. 5 controls and 7 PSA-KO mice were prepared for ex-vivo dMRI.

dMRI: Images were acquired on an actively shielded 9.4T/31cm (Agilent/Varian) system equipped with 12cm gradients coils (400mT/m) with a transceiver birdcage coil of 2.5 cm diameter. DMRI images were acquired with a spin-echo sequence with the following acquisition parameters: TE/TR=50/4000ms, resolution 140×125×300×μm³, matrix 128×96 with 36 axial slices. Diffusion scheme consists of a multi b-value shells protocol (b-value=1980, 3750 and 5630s/mm²) with constant diffusion gradient application time and separation (δ/Δ=5.5/30ms). A total of 81 non-collinear directions were acquired, which were split over the three shells (21/30/30). Two averages were acquired for the third shell, to compensate for the strong signal attenuation.

Processing: DTI and NODDI models were reconstructed using the NODDI matlab toolbox⁵. The NODDI models was adapted for ex-vivo acquisition: D_ic=0.6×10^-3mm²/s and D_iso=2.0×10^-3mm²/s and an extra compartment for water bound to fixative was added. DT images were then used to construct a study-specific DT template using DTI-TK⁶. The regions of interest (ROI) were drawn on the study-specific template; ROIs were transformed back to the subject space to compute ROI-averaged estimates from DTI and NODDI maps. From DTI: fractional anistropy (FA), mean, axial and radial diffusivities (MD, RD and RD); from NODDI, orientation dispersion index (ODI), intra-neurite, isotropic and bound volume fraction (ν_ic, ν_iso, ν_bd) metrics were tested between groups using a t-test.

The good quality and high resolution of the dMR images and models maps (Fig. 1&2), with the generation of the study-specific DT template, ensured an accurate identification of small white matter fiber tracts investigated in this study (e.g. stria-terminalis, anterior comissure). High quality registration in DT space of the subjects from DTI-TK also minimizes the inter-subject variations⁷. 7 ROIs were investigated in the current study covering major fiber tracks (corpus callosum, CC; internal capsule, IC; external capsule, EC; fornix, FX) and specific fibers controlling anxiety and social behavior (bed nucleus of the stria-terminalis, ST; posterior part of the stria-terminalis, STpost; anterior comissure, AC) shown in Figure 3.

Overall only minor changes were visible between the PSA-KO and control mice. Trends of microstructural alterations were visible in a few fiber tracts such as ST, ST-post, AC and FX (Fig. 4). The ROIs affected by the lack of PSA-NCAM during development show a reduction of FA driven by an increased RD, which was also reflected in larger MD values. The larger uncertainty in the NODDI metrics resulted in weaker trends with an increased neurite dispersion (ODI) and reduced ν_ic (Fig. 4). Both DTI and NODDI models converged to a disorganization of the fibers with a diminished density, which agrees with the known neurodevelopmental deficit of the PSA-KO mice.

The expression of PSA-NCAM, in the PSA-KO micee, is not totally suppressed and only decreased of 45% at postnatal day 1⁸, which could explain the moderate alterations reported by dMRI. Nevertheless, the tracts showing the most significant trends are part of the amygdalocortical pathway which is in good agreements with the behavioral alterations of the PSA-KO mice, which present impaired working memory, reflecting schizophrenic like phenotype as well as decreased anxiety and sociability (Fig 5).

These preliminary ex-vivo results suggest that altered plasticity during development of the PSA-KO mice, presenting schizophrenic like phenotype and altered sociability, creates long term effects and structural alterations depicted by dMRI. Despite the moderate changes reported, probably due to the low number of subject, this work motivates to pursue in-vivo study with the great advantages of the none-invasive property of dMRI offering the possibility to perform longitudinal study during developmental age.