(1)H MRS detection of glycine residue of reduced glutathione in vivo

Abstract

Glutathione (GSH) is a powerful antioxidant found inside different kinds of cells, including those of the central nervous system. Detection of GSH in the human brain using (1)H MR spectroscopy is hindered by low concentration and spectral overlap with other metabolites. Previous MRS methods focused mainly on the detection of the cysteine residue (GSH-Cys) via editing schemes. This study focuses on the detection of the glycine residue (GSH-Gly), which is overlapped by glutamate and glutamine (Glx) under physiological pH and temperature. The first goal of the study was to obtain the spectral parameters for characterization of the GSH-Gly signal under physiological conditions. The second goal was to investigate a new method of separating GSH-Gly from Glx in vivo. The characterization of the signal was carried out by utilization of numerical simulations as well as experiments over a wide range of magnetic fields (4.0-14T). The proposed separation scheme utilizes J-difference editing to quantify the Glx contribution to separate it from the GSH-Gly signal. The presented method retains 100% of the GSH-Gly signal. The overall increase in signal to noise ratio of the targeted resonance is calculated to yield a significant SNR improvement compared to previously used methods that target GSH-Cys residue. This allows shorter acquisition times for in vivo human clinical studies.

Figure 1

High resolution experimental NMR spectrum (top) of reduced glutathione (B₀=11.7 T, T = 37°C, pH=7.2) and simulations (bottom) (a). The simulations show a good agreement with an experiment (e.g. expanded GSH-Cys residue region (b)). However when the simulation of GSH-Gly is carried out using previously reported parameters, either a singlet (c) or a doublet (J=5 Hz) (d) (with the same line width used for simulations of GSH-Glu and GSH-Gly), there is a large discrepancy with the experimental data. This discrepancy appears to be not just due to a small difference in line width or J-value.

Figure 2

Experimental GSH spectra (B₀ = 9.4 T, T = 37°C, pH = 7.2) as a function of the echo time. Note the absence of J-evolution for GSH-Gly signal and that at TE=200 ms, GSH-Gly signal is significantly reduced, with most of the signal at 3.77 ppm remaining due to GSH-Glu residue.

Figure 3

(color). Expanded experimental GSH-Gly and GSH-Glu signal region and simulations at 3.77 ppm as a function of magnetic field strength (a–d) and as a function of the temperature (d–f). All the spectra were collected at pH=7.1–7.2. The spectra at higher field were acquired with pulse-acquire (11.7 and 14 T), whereas the spectra at lower field were acquired with spin-echo (4.7 and 9.4 T, echo time = 19 ms). The model for GSH-Gly was based on the two singlet peaks separated by 3.4 Hz (T = 37°C). The additional line broadening (Gaussian LB = 100 ms) due to the loss of the coherence during the exchange process is used to simulate GSH-Gly peaks for all four field strengths (4.7 – 14 T). After a small temperature decrease (~7°C), the GSH-Gly pattern shows distinct separation of the glycine peaks (~5.3 Hz) (e), and at room temperature the separation is increased further (~ 6.5 Hz) (f).

Figure 4

Demonstration of the J-difference editing method for separating GSH-Gly from Glu and Gln. The editing pulse is applied at 2.1 ppm for the edit on scan (a), which refocuses Glu and Gln resonances at 3.77 ppm. In the edit off scan (b) Glu and Gln resonances undergo non-refocused J-evolution. In both edit on and edit off scans, GSH-Gly resonance is not affected. The difference spectrum (c) is fitted to extract Glu and Gln contribution, which is subsequently used as a constraint in the fitting of the summed spectrum (d). The individual contributions of Glu, GSH-Gly and Gln were simulated with 5:1:1 concentration ratios (with line broadening factors of 150 ms and 200 ms for Lorentz and Gauss, respectively and additional Gauss line broadening of 100 ms for GSH-Gly). In vivo spectra are shown without using apodization.

Figure 5

Comparison of the J-difference editing methods for GSH-Gly and GSH-Cys residues: (a) The simulation of GSH-Cys (difference spectrum, MEGA-PRESS, TE=68 ms, exponential LB = 45 ms) vs GSH-Gly signal (summed spectrum, PRESS+4, TE=72 ms, exponential LB= 45 ms, Gaussian LB = 100 ms) at 4 T. In vivo spectra were collected in the precuneus brain region of a healthy volunteer (b). The editing of cysteine residue using J-difference MEGA-PRESS is demonstrated in (c). PRESS+4 editing results are shown in (d). The difference spectrum contains only the contribution from Glx at 3.77 ppm (co-edited GABA signal is also present at 3 ppm). This contribution calculated from the difference spectrum is used as a constraint in the fitting of the summed spectrum. The residual summed spectrum (dotted line) is shown after subtracting MM and Glx.