Purpose

The inhibitory/excitatory balance between neurotransmitters GABA and glutamate, is fundamental to the homeostatic regulation of neuronal firing (1,2). Proton magnetic resonance spectroscopy (¹H-MRS) is capable of measuring concentrations of both metabolites, and while the molecular behavior of these neurotransmitters is known, relationship between their MRS signals, neural signaling and behavior is not fully understood. Even less is known about the link between neurotransmitters and hemodynamic responses. While baseline GABA concentrations have been related to CMRO₂ estimations (3), the relationship between task GABA concentrations and CMRO₂ has yet to be examined. Behavior of the GABA+ and glutamate+glutamine (Glx) ¹H MRS signals under visual stimulation is examined here and related to hemodynamic imaging contrasts in the human visual cortex.

Methods

Experiments were performed on a Philips Achieva 3T scanner. Editing of GABA+ (GABA+macromolecules) was performed using a single-voxel MC MEGA sequence (4) in 6 healthy volunteers (4 female, age 27.1±2.7 years) in a 3 cm³ voxel in the Occipital lobe (OCC) with local Institutional Review Board approval. Acquisition parameters include 14-ms editing pulses at 1.9-ppm and 7.5-ppm in edit-ON and edit-OFF , respectively, TE/TR=68ms/2.63s, 448 transients, spectral bandwidth=2-kHz, and scan duration=20 minutes. An alternating blue/yellow flashing checkerboard (27/54s rest/stimulus×16) was used (figure 1a). Two separate edited-MRS scans were performed at different flashing frequencies: 8Hz and 4Hz with a 20-minute break in between. MRS results reflect pooled data from both frequencies. Data was processed using in-house software and Gannet 3.1 (5) with time-domain retrospective phase and frequency correction (6). As the full water signal is made available at each transient by metabolite-cycling, differences in the average water linewidth between rest and stimulation were extracted to estimate the BOLD response and expressed as percentages of the water linewidth during rest. Edited spectra were linewidth-corrected based on these values, spline-baseline-corrected, and fit with Gannet.

CMRO₂ was measured using the Fick principle as described previously (7-9). Briefly, CMRO₂ (µmol O₂/min/100g brain-tissue) was quantified based on arterio-venous difference in oxygen content (10) and is written as CMRO₂ = CBFxOEFxCa = CBFx(Ya – Yv)xCa (1), where CBF (cerebral-blood-flow) represents the blood supply to the brain, Ya and Yv represent the arterial and venous blood oxygen saturation fraction (in %) respectively, and OEF= (Ya - Yv) is the oxygen-extraction-fraction. Ca represents the oxygen carrying capacity of unit volume blood and is 8.97 mmol O₂/mL blood (11). CBF was measured at rest, during 4Hz and 8Hz checkerboard stimulation using a pseudo-continuous-arterial-spin-labeling (pCASL) sequence with following parameters: FOV=220×220mm², matrix=64×64, 29 axial slices, 5mm-thick, TR/TE=4750/16-ms, label-duration=1.8s, post-label-delay=1.8s, single-shot echo-planar-imaging (EPI), 30 label/control image pairs, 2D readout without background suppression, and duration=4-minutes 45s. Ya was measured using a pulse oximeter and Yv was measured non-invasively in the sagittal sinus using TRUST-MRI (12). OCC (MNI template-space) CBF was used to calculate OCC CMRO₂.

Results

Figure 1b shows the water linewidth MRS time-course in one subject. As expected, the water linewidth decreases by a median of 0.11Hz with stimulation relative to rest, consistent with literature (13-15). Representative GABA-edited spectra show changes in GABA+ and Glx between rest and stimulation (figure 1c). Correlations between changes in GABA+ and Glx concentrations between stimulation and rest, and GABA+ and Glx concentrations at rest versus the BOLD response are plotted in figure 2. Both GABA+ (figure 2a) and Glx (figure 2b) concentration changes display a strong positive correlation with the BOLD response, while resting GABA+ (figure 2c) and Glx (figure 2d) levels show trend to a negative correlation. Confirming these results, GABA+ (figure 3a) and Glx (figure 3b) changes during stimulation were significantly negatively correlated with GABA+ and Glx concentrations at rest, respectively. GABA+ and Glx concentration changes during stimulation were also positively correlated with one another (figure 3c). While there was no statistically significant difference in CBF between stimulation and rest (figure 4a), CMRO₂ was higher with 8Hz stimulation than at rest (figure 4b). Additionally, task GABA+ concentrations are negatively correlated with CMRO₂ changes (figure 4c).

Discussion

Here, it’s shown that MC MEGA can measure dynamic changes in GABA+ and Glx concentrations during visual stimulation. GABA+ and Glx changes are positively correlated with the BOLD response and GABA+ increases are accompanied by Glx increases indicating that excitatory processes might be activated during stimulation as well as compensatory inhibitory processes to maintain the inhibitory-excitatory balance. Alternatively, Glx reflects changes in both glutamate and glutamine, and most of glutamate is a product of the TCA cycle reflecting an increase in oxidative energy metabolism. In line with previous work, glutamate changes could reflect a shift in metabolism to a higher state in response to prolonged stimulation (15-17). It is also shown that GABA+ and Glx increases are negatively correlated with their resting concentrations. This is possibly due to differences in resting GABA+ levels among individuals and consequently differences in effect sizes of GABA+ and Glx upregulation in response to stimulation (18). Also, GABA+ levels during stimulation are negatively correlated with CMRO₂ changes, which suggests that an alternate faster energy source from glucose during glycolysis (18-21) may have supplemented energy requirements during prolonged visual stimulation. Consequently, higher task GABA+ levels indicate more inhibitory neurotransmission, which may be neuroprotective, preventing the brain from exhaustion.

Acknowledgements

This work was funded by the Cancer Prevention and Research Institute of Texas (CPRIT) (Grant Number: RR180056).