The underlying biophysical mechanism of frequency shifts due to changes in white and gray matter (WM, GM) tissue microstructure have been an area of increased focus for assessment of injury and disease pathology in brain[1] and spinal cord[2]. Sources of this contrast have been attributed to factors such as iron, local bulk tissue magnetic susceptibility, and local tissue microstructure (fiber orientation)[3-5]. The rat spinal cord possesses well-defined anisotropic WM tissue architecture that produces predictable patterns of degeneration[6]. Compared to myelin water fraction (MWF) and diffusion imaging (DTI), frequency shifts have shown higher sensitivity to pathology in such ex-vivo rat injury models[7]. This study has incorporated focal demyelination in cervical rat spinal cord using lysolecithin injections into the dorsal column, which demyelinates WM while leaving axons intact[8]. Our results show frequency shift evolution in the dorsal column losing contrast with surrounding gray matter as demyelination progresses. Furthermore, past 14 days, WM-GM contrast is restored as the effects of lysolecithin declines and remyelination occurs.

Lysolecithin: Following laminectomy, 0.5μL of 1% lysolecithin (LPC, Sigma, St. Louis, MO) was injected 0.7mm deep into the C5) dorsal column of healthy female SD rat spinal cords[11]. Injection was performed over 20 minutes to prevent backflow. Imaging time-points at 1,3,5,7,14,28, and 42-day post-injection. Spinal cords excised at set end-points, perfused, and paraformaldehyde-fixed for ex-vivo imaging.

Imaging: in-vivo animals were anesthetized and scanned at 7T (100μm×100μm×1000μm, TE=3.7+3ms, Echoe =4, TR=35.8ms, FA=11°, NA=20). ex-vivo cords were scanned as in [2].

Frequency Maps: MR phase images unwrapped and high pass filtered (homodyne filter) and converted into frequency maps [1]. Frequency maps echoes were averaged to increase overall SNR.

Histology: Post-fixation, cords transferred to 24% sucrose, cryoprotected, and cut to 20μm-thick cross-sections. Sections were stained and imaged (detailed in [9]).

Studies have found strong frequency shift correlations with demyelination and axonal damage histology markers in rat spinal cord injury models[2] similar to MR behaviour predicted for brain WM diseases[10]. Because damage to myelin and axons generate either positive or negative (respectively) shifts in frequency, specific contributions from each source is difficult to quantify independently. The lysolecithin model has been shown in brain to produce demyelination while preserving axon integrity, allowing assessment to focus on myelin changes over time[12]. Using superfine glass micropipettes, we ensure mechanical injury is minimal and localized to a sub-mm area, as seen in Fig.1, which also shows excellent WM/GM contrast in-vivo. Demyelination is expected to occur through days 5-14[8], where we observe WM/GM contrast reduction in frequency shifts (but not magnitude) [Fig.2]. The WM/GM contrast returns at day 28, consistent with similar studies from literature[8]. Tissue pathology was confirmed with immunohistochemistry. The first two row in Fig.3 show sections where clear axonal loss (Nf/TubII stain) is seen at the injection site at day 1. In row two, the sections from the same animal +1mm cranially show almost no axonal loss. While no demyelination is observed day 1, both degen-MBP and myelin stains show strong demyelination around the injection site and needle path at days 5 and 7 due to the lysolecithin. Quantification of in-vivo scans (Fig.4, left) show frequency shifts in the dorsal column progressively increasing, becoming GM-like. At Day 14, the average frequency increases to its peak before returning to levels more consistent with myelinated white matter at days 28, 42. Similarly, ex-vivo scans (Fig.4, right) exhibited a trend of increasing frequency shift at days 5, 7, and peaking at day 14, followed by recovery of WM/GM contrast at day 42. However, this trend is not as clear as in-vivo due to limited cords available: n=2 per timepoint, and n=1 at day 42, and additional cords that are still being tracked in-vivo have not been included yet.In this study, lysolecithin demyelination in rats produced a mode of myelin injury and repair that minimally affected axonal integrity, allowing assessment of predominantly myelin-related pathology changes with frequency shift mapping both in-vivo and ex-vivo. We show strong frequency shift sensitivity to the injury longitudinally that was not apparent in magnitude images, and put forward an initial prediction of the frequency shift evolution over periods of progressive demyelination (days 1 through 14) and remyelination (days 28, 42). Additionally, though not shown, diffusion images were acquired in-vivo and ex-vivo for additional confirmation of axonal integrity, and harvested cords are saved for histological quantification as well. Further assessment of the lysolecithin model will be beneficial for applications of frequency shift mapping to studies across a wide variety of WM-diseases, especially demyelinating diseases such as multiple sclerosis and neuromyelitis optica.