Early detection and monitoring of cerebral ischemia using calcium-responsive MRI probes

Significance The duration of cerebral ischemia is a key factor in determining the severity of brain damage and the course of action. Thus, an accurate and timely observation of the ischemic process is highly critical. Here we present a molecular neuroimaging approach that enables direct detection and real-time visualization of transient cerebral ischemia. The method relies on high-resolution observation of extracellular calcium alterations associated with the spatiotemporal dynamics of cerebral ischemia, using a selective molecular MRI probe. The rapid detection of calcium fluctuations in healthy and disease states will not only lead to essential insights for successful treatment and recovery of ischemic brain tissue but will also improve our understanding of the underlying neurobiology of neurological and psychiatric disorders.


Supplementary Methods
Preparation of Gd 2 L 1 and Gd 2 L 2 . Commercially available reagents and solvents were used without further purification. Purification by column chromatography was performed with silica gel 60 (0.03−0.2 mm) from Carl Roth (Germany). 1 H and 13 C NMR spectra were recorded on a Bruker Avance III 300 MHz spectrometer at 25 °C. High resolution mass spectra were recorded on a Bruker Daltonics APEX II (FT-ICR-MS) with an electrospray ionization source. Gd 2 L 1 (Ca-responsive contrast agent) was prepared as previously reported (1). The compound L 2 (non-responsive contrast agent) was prepared according to the reaction scheme presented in the Supplementary Fig. 1. Compound 1 was prepared as previously reported (2). Compound 2 (0.31 g, 0.82 mmol) was dissolved in acetonitrile (20 mL) in a 50 mL round bottom flask. Potassium carbonate (0.45 g, 3.28 mmol) was then added as a solid to the stirring mixture, followed by a solution of 1 (1.6 g, 1.97 mmol) in acetonitrile (10 mL).
The reaction mixture was stirred for 16 hours at 60 °C. The resulting mixture was cooled, the inorganic salts were removed by filtration and the solution was evaporated in vacuo. The crude product was purified by column chromatography (silicagel, 0-15% MeOH in CH 2 Cl 2 ) to obtain pure 3 as a yellow oil (530 mg, 35% yield). 1   was applied to the normalized signals (6). K-means algorithm requires the number of clusters as an input. In order to ensure the robustness of the analysis with respect to the choice of parameter K, clustering was conducted for K ranging from 2 to 10 for all datasets and the S5 results were statistically compared. Thereby, K=2 presented the optimal choice. The regions of interest (ROIs) were then defined using a hierarchical clustering algorithm.

Complexation of L 2 with Gd
The algorithm scheme is as follows: 1.
Apply the K-means clustering algorithm with two clusters to all the voxels in the initial mask; 2. Select all the voxels corresponding to the centroid with a larger mean value;

3.
Apply K-means with two clusters to the voxels selected in step 2;

4.
Choose the voxels corresponding to the centroid with a larger mean value, and use them as our ROI.
The critical advantage of this algorithm is that it identifies ROIs systematically and without a priori assumptions on the number of voxels or total mask size. Here, the algorithm leads to Thereby, the spline function f sp (t) is extrapolated over segment 2 assuming as if it follows the same trend as is inherent in segments 1 and 3.
For non-ischemic experiments, the detrending procedure is identical to the above algorithm, with the difference that step 3 (ignore) is not performed. The spline function with segment 2 is then based on data from the experiment and not, as in case of ischemia, based on the obtained trends from segments 1 and 3.