Chemical Exchange Saturation Transfer (CEST) Signal at -1.6 ppm and Its Application for Imaging a C6 Glioma Model

Abstract

The chemical exchange saturation transfer (CEST) signal at -1.6 ppm is attributed to the choline methyl on phosphatidylcholines and results from the relayed nuclear Overhauser effect (rNOE), that is, rNOE(-1.6). The formation of rNOE(-1.6) involving the cholesterol hydroxyl is shown in liposome models. We aimed to confirm the correlation between cholesterol content and rNOE(-1.6) in cell cultures, tissues, and animals. C57BL/6 mice (N = 9) bearing the C6 glioma tumor were imaged in a 7 T MRI scanner, and their rNOE(-1.6) images were cross-validated through cholesterol staining with filipin. Cholesterol quantification was obtained using an 18.8-T NMR spectrometer from the lipid extracts of the brain tissues from another group of mice (N = 3). The cholesterol content in the cultured cells was manipulated using methyl-β-cyclodextrin and a complex of cholesterol and methyl-β-cyclodextrin. The rNOE(-1.6) of the cell homogenates and their cholesterol levels were measured using a 9.4-T NMR spectrometer. The rNOE(-1.6) signal is hypointense in the C6 tumors of mice, which matches the filipin staining results, suggesting that their tumor region is cholesterol deficient. The tissue extracts also indicate less cholesterol and phosphatidylcholine contents in tumors than in normal brain tissues. The amplitude of rNOE(-1.6) is positively correlated with the cholesterol concentration in the cholesterol-manipulated cell cultures. Our results indicate that the cholesterol dependence of rNOE(-1.6) occurs in cell cultures and solid tumors of C6 glioma. Furthermore, when the concentration of phosphatidylcholine is carefully considered, rNOE(-1.6) can be developed as a cholesterol-weighted imaging technique.

Keywords: chemical exchange saturation transfer (CEST); cholesterol; glioma; magnetic resonance imaging (MRI); relayed nuclear Overhauser effect (rNOE).

Figure 1

Average z-spectra (N = 9) from normal (blue) and tumorous (red) brain as measured on days 3 (A), 6 (B), 9 (C), and 12 (D). Shading areas indicate standard deviation resulting from animal variation. Saturation amplitude and saturation duration are 0.9 μT and 4 s, respectively.

Figure 2

Expansion of z-spectra from Figure 1 for comparative purposes (A–D); the corresponding residuals of apparent exchange-dependent relaxation rate (AREX_resid) are plotted from (E–H), which include averaged signals (N = 9) from normal (red) and tumorous (blue) brain tissues and their standard deviations (shaded areas). Averaged AREX_resid signals from (E–H) are deconvoluted using a Gaussian function centered at approximately −3.5 ppm (I–L), in which dashed lines indicate rNOE(−3.5) and solid lines indicate residues from deconvolutions.

Figure 3

Longitudinal changes in magnitudes of AREX_resid at −3.5 ppm (A) and −1.6 ppm (B), as extracted from Figure 2 (N = 9); longitudinal changes in area under the curve (AUC) of corresponding AREX_resid at −3.5 ppm (C) and −1.6 ppm (D). (* p < 0.05, ** p < 0.001).

Figure 4

Images of AREX_resid(−1.6) (A–D), hematoxylin staining (E–H), and filipin staining (I–L) on days 3, 6, 9, and 12. The arrows indicate the tumor regions.

Figure 5

¹H NMR of cholesterol methyl (0.68 ppm) from lipid extracts of normal and tumorous brain tissues (N = 3) on days nine (A) and 12 (B) and their cholesterol content (C) as quantified using AUCs of methyl shown in (A,B). (D) The choline methyls from the phosphatidylcholines (~3.35 ppm) and alkyl hydrogens (~1.5 ppm) can be identified in the ¹H spectra. The AUCs between 3.28 to 3.38 ppm and 1.45 to 1.55 ppm were used to estimate the relative amounts of phosphatidylcholine (E) and total lipid (F). (* p < 0.05).

Figure 6

Averaged (N = 3) z-spectra (A,D) from cholesterol-depleted (red), control (green), and cholesterol-enriched (blue) cell homogenates, corresponding AREX_resid (B,E), and deconvoluted AREX_resid (C,F). Standard deviations are indicated by shaded areas. Dotted lines indicate rNOE(−3.5), solid lines indicate rNOE(−1.6), and dashed lines indicate residues from deconvolutions. The CEST signals from the negative and positive regions are shown in (A–C) and (D–F), respectively.

Figure 7

(A) Averaged ¹H NMR signal (N = 3) of cholesterol methyl at 0.68 ppm from cholesterol-depleted (red), control (green), and cholesterol-enriched (blue) cell homogenates, in which standard deviations are indicated by shaded areas. Scattering plot illustrating the association between cholesterol concentration and AUCs of AREX_resid(−3.5) (B) and AREX_resid(−1.6) (C), in which regression is indicated by a blue line.