Metabolic analyses have revealed that cancer orchestrates the use of glucose (via glycolysis or Warburg effect) and glutamine to meet the demand of energy and building blocks necessary for survival and proliferation.^{1, 2} In contrast to normal cells, many cancers extensively use the amino acid glutamine via glutaminolysis pathway.^{4, 5} Mitochondrial enzyme glutaminase (GLS), which converts glutamine to glutamate (Glu) is the first and rate-limiting step and inhibition of GLS leads to over 2 fold change of tumor Glu concentration. Emerging therapeutics targeting cancer GLS are currently in clinical trials of cancers including the triple negative cancers, which have few therapeutic options but are highly dependent on glutamine for survival and growth. Non-invasive and sensitive method that can assess the pharmacodynamics (PD) of this new category of drugs is needed to facilitate their development and clinical implementation. Based on our prior development of chemical exchange saturation transfer (CEST) MRI to map the brain Glu concentration, here we propose to examine the utility of GluCEST ⁶ for monitoring the change of tumor Glu level in response to GLS inhibition.All the experiments were performed under an approved IACUC protocol of the University of Pennsylvania. Human triple negative breast cancer cells (HCC1806) are inoculated subcutaneously in the flank of athymic nu/nu mice and xenografts were grown to about 500 mm³ before assigned to CB-839 or Vehicle (VHE) group. A selective, potent and reversible GLS inhibitor, CB-839,⁷ was obtained from Calithera via material transfer agreement. CB-839, dissolved in a vehicle (25% (w/v) hydroxypropyl-β-cyclodextrin in 10 mmol/L citrate, pH 2) was administered orally (200 mg/kg) twice daily, while control mice received the same volume of vehicle solution. A total of four doses are administered. Each mouse underwent the GluCEST MRI before treatment and 4 hours after the last dose. All MRI studies are performed using custom-built slot-tube resonator (15 mm dia) tuned to ¹H on a 9.4T horizontal bore spectrometer. GluCEST MRI was performed using a custom-programmed RF spoiled gradient echo readout pulse sequence in 10 mm slices with a frequency selective continuous wave saturation preparation pulse. CEST images were collected using a 1 second saturation pulse at peak B1 of 250 Hz for the frequencies ± 2.5-3.5 ppm from water resonance with a step size of 0.25 ppm. The B0-corrected images at 3 ppm (M_+3ppm) and -3 ppm (M_−3ppm) were used for computing the percent GluCEST asymmetry value, which is equal to 100×[(M_−3ppm − M_+3ppm)/M_−3ppm]. Regions of interest were manually segmented from T2-weighted images.By utilizing a slot-tube RF resonator in which tumor is relatively isolated from the breathing motion, reliable B₀ and B₁ homogeneity maps were obtained in our studies (Fig 1A-B). Regions which had B₀ variation beyond ±0.3 ppm or relative B₁ ­below 0.8 or above 1.2 were excluded for analyses. Pre- and post-treatment GluCEST map from a mouse in CB-839 group is shown in Fig 1C, revealing an overall decrease of GluCEST value with heterogeneity across the tumor after treatment (bottom vs. upper panel). Our data show a marked reduction of GluCEST value in 3 out of 4 CB-839 treated mice while there is minimal change in all VEH treated mice (Fig 2). Compared to average 0.32% increase of GluCEST in VEH group (n=4), GluCEST value decreases 3.5% in CB-839 group (n=4), consistent with extract data (below) since about 1.7% GluCEST change per millimolar Glu was calibrated in phantom and brain.^6,8 Using high resolution ¹H MRS of perchloric acid extraction of the tumors, we confirmed that tumor Glu level in CB-839 treated mice was reduced over 2 mM compared to VEH treated ones, meanwhile glutamine was increased and alanine (Ala) unchanged (Fig 3). Taking together, these data corroborate a reduction of tumor GluCEST value with a decrease in Glu concentrations by 1H MRS, strongly supporting the concept of GluCEST as a marker of GLS inhibition. The presence of Ala, which also exhibits GluCEST effect, could contribute to baseline GluCEST value. It is noted that cellular Glu and Ala pool are linked via the alanine transferase (ALT) reaction (Fig 4), hence upon GLS inhibition, Ala concentration is also expected to decrease as Glu, leading to a serendipitous increase of GluCEST sensitivity. Tumor intracellular pH (pH_i) also contributes to GluCEST value, however a pH_i drop following the treatment is expected to cause an underestimation of the change in GluCEST value and we will examine it further. In summary, our preliminary results presented here suggest GluCEST is a promising PD marker for a potent glutaminase inhibitor.RZ and PB are equal contribution.