The entire experimental process is shown in Fig. 1. The Ethics Committee on Animal Study of Lanzhou University Second Hospital (Lanzhou, China) approved the protocol (approval no. D2025-787). C57 mice (weight, 17–18 g) were purchased from Experimental Animal Center of Lanzhou University. Mice were housed in the specific-pathogen-free animal care facility of Lanzhou University Second Hospital under a 12/12-h light/dark cycle (lights on at 7:00 AM), an ambient temperature of 22±2°C and a relative humidity of 50±10%. All mice were fed a standard diet and had ad libitum access to autoclaved drinking water. The animals were housed (3–5 mice/cage) in standard ventilated cages with corn cob bedding. A total of 18 male C57BL/6 mice (age, 8 weeks) were included (12 for MRI, with 6 assigned to the tumor group and 6 to the normal group; 6 for PET-CT, with 3 assigned to the tumor group and 3 to the normal group).

In the surgery, mice were anesthetized for 5 min using 3% isoflurane (2 l/min). Subsequently, anesthesia was maintained at 1–1.5% isoflurane (1 l/min) until the procedure was completed. The mice were placed on a thermal pad, and their heads were secured in a stereotaxic instrument. After the head was sterilized with iodophor, the skin was cut longitudinally in accordance with the sagittal midline position; the length of the incision was ~1 cm, and the position corresponding to the right caudate nucleus was determined according to the anatomical map of the stereotaxic instrument: 2.0 mm anterior to the fontanel, 1.0 mm right to the midline and 3.5 mm subdural in depth. A hole was drilled with a skull drill without damaging the dura mater. A total of 5 µl (5×10⁴ GL261 cells) cell suspension was injected using a glass electrode at an injection rate of 50 nl/sec. After the cell suspension was injected, the needle was retained for 10 min. The incision was sutured with a 5-gauge thread and sterilized with iodophor to prevent infection. Following surgery, mice were kept on a heating pad until full recovery from anesthesia and monitored closely for the first 6 h. Health and behavioral assessments were conducted every 2 days initially, with the frequency increasing to daily from day 14 onward. Humane endpoints requiring immediate euthanasia included weight loss >20% of peak body weight. No mice died during the 21 day experimental period following the cell suspension injection. All mice were euthanized at the predetermined experimental endpoint for tissue sampling. Death was confirmed by cessation of chest movement and spontaneous breathing, and the absence of a palpable heartbeat for ≥2 min.

The murine glioma GL261 cell line was obtained from Zhili Zhongte Biological Technology Co., Ltd. Cells were cultured in DMEM, supplemented with 10% fetal bovine serum (both Beijing Solarbio Technology Co., Ltd.,), and 100 µg/ml streptomycin. All cells were maintained at 37°C in a humidified incubator with an atmosphere of 5% CO₂.

GL261 cells were injected into the right caudate nucleus of the mice to establish the in situ brain tumor model. A total of 12 mice underwent regular in vivo scans using MRI: T2WI was performed on days 7, 14 and 21 after implantation to non-invasively evaluate the growth of the tumor tissue. Additionally, ¹H-MRS was used to measure brain metabolites in mice 21 days after tumor implantation. Mice were anesthetized with 3% isoflurane at 2 l/min for up to 5 min, followed by 1–1.5% isoflurane at 1 l/min consistently during the MRI scan. All MRI and MRS experiments were acquired on the horizontal 30 cm-bore preclinical 9.4T MR system (uMR 9.4T, Shanghai United Imaging Healthcare Co., Ltd.) using a volume coil with 86 mm inner diameter for radio frequency transmission and a three-channel phased array mouse brain surface coil for signal reception. To stabilize the body temperature of the mice, an animal warming system was used, which consisted of a warm water (39°C) reservoir with a pump and hoses underneath the animal bed. The mice breathed freely during the whole MR acquisition and were monitored for changes in respiratory rate to adjust the anesthetic concentration. Body temperature and respiratory rate were kept at 36–37°C and 60–80 bpm, respectively, using small animal vital signs monitor system (SA instruments, Inc.) during the MRI scan.

A fat suppressed T2-weighted fast spin echo sequence was performed for transverse imaging of the mouse brain to evaluate brain tumor, with the following parameters: Repetition time, 3,000 msec; echo time, 49 msec; volume of interest (VOI), 20×20 mm²; resolution, 0.1×0.1 mm; echo spacing, 7.04 msec; echo train length, 13; slice thickness, 0.5 mm with no gap. Imaris software (version 10.2; Oxford Instruments) was used for 3D reconstruction as well as rendering of mouse brain tumors.

Water-suppressed and non-water-suppressed MRS data of mouse brain tumor and the contralateral brain tissue were acquired using the Point Resolved Spectroscopy sequence with following parameters: Repetition time, 2,500 msec; echo time, 6 msec; VOI, 2×2×2 mm³; bandwidth, 4,000 Hz; number of excitations (NEX), 128. After positioning of the VOI, manual shimming (defined as the manual optimization of magnetic field homogeneity within each VOI) was adjusted for each VOI. To avoid residual water signals in the spectral data caused by suboptimal shimming results, Java Based Magnetic Resonance User Interface software (version 5.2) (26) was used to suppress the water signal in the MRS data (27,28). Metabolite quantification was performed using the LCModel (version 6.3-1L) (29,30), which calculates the best fit to the experimental spectrum as a linear combination of model spectra (simulated spectra of brain metabolites). Raw data were used for the standard data input. The water-suppressed time domain data were analyzed between 0.2 and 4.0 ppm. The following metabolites were included in the basis set of simulated metabolite spectra used to model and quantify the in vivo MRS data: creatine (Cr), glycerophosphorylcholine (GPC), phosphorylcholine (PCh), glutathione, myo-inositol (mI), lactate (Lac), phosphocreatine (PCr), N-acetylaspartate (NAA), N-acetylaspartylglutamate (NAAG) and lipids at 1.3 ppm (Lip1.3). The absolute quantitative concentrations of metabolites in the mouse brain were calculated using the LCModel software, with reference to the internal tissue water signal. Key metabolite ratios were calculated, including tNAA/tCr, tCho/tCr, tNAA/tCho, Lip1.3/tCr, Lac/tCr and mI/tCr, where tNAA=NAA + NAAG, tCr=Cr + PCr and tCho=GPC + PCh. Additionally, the composite lipid peak at 1.3 ppm was analyzed, defined as Lip1.3=Lip1.3a (at ~1.28 ppm) + Lip1.3b (at ~1.30 ppm). The ¹H-MRS dataset analyzed during the present study is available in the Zenodo repository (zenodo.org/records/17759344) at the following DOI: 10.5281/zenodo.17759344.

Following induction with 3% isoflurane (2 l/min for up to 5 min), mice were maintained under anesthesia with 1.0–1.5% isoflurane at 1 l/min during a 2 h, whole-body FDG dynamic PET/CT scan (MadicLab PSA071, Shandong Madic Technology Co., Ltd.). At the start of the scan, mice were injected via a tail vein indwelling needle with 100 µl FDG solution, containing 8.4±0.7 MBq radioactivity. PET/CT images were reconstructed in three-dimensional reconstruction algorithm for maximum likelihood analysis (3D RAMLA; MadicLab PSA071, Shandong Madic Technology Co., Ltd.) with CT scan (80 kV, 70 mAs) for attenuation correction and fusion localization, with 0.8×0.8×0.8 mm as the final resolution. PMOD software (version 4.4, PMOD Technology GmbH) was used for data processing and image analysis. For visual assessment of tracer distribution, maximum intensity projection images were generated. Based on PET/CT imaging, VOI of the liver, kidney and bladder was manually plotted. The VOI of the tumor and the corresponding anatomically matched region on the contralateral hemisphere (serving as the background) were outlined as a sphere with a diameter of 1.5 mm. The standardized uptake value of FDG in the VOI (SUVmax) was calculated. The tumor/background (T/B) ratio was calculated as SUVmax-T/SUVmax-B. Time-activity curves (TACs) were extracted from organ VOIs.

A 2-tissue compartment model (plasma and precursor and metabolic product pool in brain tissue) was constructed (Fig. 2). k1 (ml/min/cm³) and k2 (min⁻¹) represent the blood-to-tissue and tissue-to-blood FDG delivery rate, respectively; k3 (min⁻¹) is the FDG phosphorylation rate. This irreversible model assumes negligible FDG dephosphorylation [FDG dephosphorylation rate (k4; min⁻¹) is 0] (31). In addition, VOI placement and TAC extraction were performed for the inferior vena cava to obtain the image-derived input function. K_i was calculated to represent the net FDG influx rate, which is used to describe glucose (32).

The mice were euthanized by an overdose of 5% isoflurane delivered at 2 l/min for 5 min and perfused transcardially with cold physiological saline. The brain was dissected and tissue was immediately collected on ice and stored at −80°C. Total protein was extracted using RIPA lysis buffer (Beyotime Institute of Biotechnology; cat. no. P0013B) supplemented with PMSF. BCA protein assay kit was used to quantify the protein concentration. The protein sample (20 µg/lane) was separated by 10% SDS-PAGE and transferred to a PVDF membrane. The membranes were blocked with 5% BSA (Wuhan Servicebio Technology; cat. no. G2052) for 1 h at room temperature. Subsequently, they were incubated overnight at 4°C with the following primary antibodies: Glucose transporter type 1 (GLUT1, (Proteintech Group, Inc.; cat. no. 66290-1-Ig, 1:1,500) and GAPDH (Wuhan Servicebio Technology Co., Ltd.; cat. no. GB12002, 1:2,500). The next day, membranes were washed three times (10 min each) with TBST (0.1% Tween-20) and incubated with HRP-conjugated secondary antibodies (Wuhan Servicebio Technology; cat. no. GB23301, 1:5,000) at room temperature for 1 h. The membranes were washed again three times with TBST (10 min each). Protein bands were visualized using ECL reagent (Biosharp Biotechnology) and imaged with a gel imaging system. The relative gray values of the protein bands were analyzed using ImageJ (version 1.50b) software.

The mice were euthanized by an overdose of 5% isoflurane delivered at 2 l/min for 5 min, perfused transcardially with cold physiological saline. The dissected brain was fixed in 4% PFA at 4°C for 24 h. Following fixation, tissue underwent graded dehydration, cleared in xylene, embedded in paraffin, and sectioned at a thickness of 5 µm. For hematoxylin and eosin (HE) staining, sections were deparaffinized, rehydrated stained with hematoxylin (5–8 min) and eosin (1–3 min) at room temperature, dehydrated, cleared, and mounted. Stained sections were examined and imaged under a light microscope for morphological analysis.

All data are presented as the mean ± SD of ≥3 independent biological replicates unless otherwise specified. Statistical analysis was performed using Prism 9 (GraphPad Software, Inc.; Dotmatics). For comparisons between two groups, a paired Student's t-test was used. For comparisons among >2 groups, one-way ANOVA was performed, followed by Tukey's post hoc test for multiple comparisons. P<0.05 was considered to indicate a statistically significant difference.

A small animal 9.4T MRI was used to evaluate whether the GBM tumor model in the mouse brain was successfully established. The tumor was observed longitudinally 7, 14 and 21 days after injection of GL261 cells in the brain under T2WI condition. Compared with normal C57BL/6 mouse MRI (Fig. 3A), the tumor was located in the right cerebral hemisphere, with a spherical shape, relatively clear boundary between the tumor tissue and the surrounding normal brain tissue, non-uniform signal within the tumor and a small number of edema bands around the tumor (Fig. 3B-D). The midline of the brain was shifted to the left, the tumor occupying effect was notable and the right lateral ventricle had different degrees of compression and deformation (Data S1, Data S2, Data S3, Data S4). The tumor grew in a time-dependent manner, measuring 1.26±0.65, 4.92±0.55 and 13.85±2.83 mm³ on day 7, 14 and 21 respectively (Fig. 3E; Table SI).

The metabolism in the tumor was different from that of the contralateral normal brain tissue as well as normal C57 mouse brain tissue. The levels of tCho (Fig. 4A) and the tCho/tCr ratio (Fig. 4E) were significantly increased in the tumor compared with that in the contralateral normal brain and the normal mouse brain tissue. Levels of tNAA (Fig. 4C) and tCr (Fig. 4B), key brain metabolites, were decreased in tumor compared with normal brain tissue. A notable decline in tNAA/tCho (Fig. 4F) ratio and increase in tNAA/tCr (Fig. 4D) ratio were demonstrated in the tumor compared with the contralateral and the normal mouse brain tissue. Lip1.3/tCr (Fig. 4G), Lac/tCr (Fig. 4H) and mI/tCr (Fig. 4I) in the tumor were significantly higher than in the contralateral normal brain tissue and the normal mouse brain tissue.

The diagnosis of tumors was confirmed using small animal PET-CT 21 days following intracerebral injection of GL261 cells in mice. FDG was rapidly distributed throughout the body of the mice after injection via the tail vein, but in the brain on the tumor side, FDG was absent for the first 6 min, after which the FDG content in the tumor tissue increased (Fig. 5F). By contrast, the FDG content in normal brain tissue decreased in a time-dependent manner (Fig. 5E). Calculation of T/B revealed a time-dependent increase (Fig. 5C), which was significantly higher at 2 h (T/B=2.23) compared with the conventional time point imaging modality of 1 h after FDG injection (T/B=1.55; Fig. 5D). SUV decreased in the kidney and liver (Fig. 5G-J), whereas SUV increased in the bladder in a time-dependent manner (Fig. 5H and L). FDG was specifically and selectively distributed in tumor tissue rather than in normal brain tissue and it was easier to differentiate tumor from normal brain tissue after extending the scan time to 2 h (Fig. 5A and B; Data S5 and S6).

Pharmacokinetic analysis showed that the distribution of FDG in brain and tumor tissue was consistent with a two-compartment model. The metabolism of FDG was abnormal in the tumor compared with brain tissue in GBM mice, represented by decreased uptake of FDG and slow metabolism of FDG in the tumor tissue (k1=0.797 vs. 1.844 for the uptake, k2=2.722 vs. 3.844 for the metabolism, k3=0.319 vs. 0.280 for the phosphorylation, νB=2.3% vs. 3.6% for blood volume fraction, K_i=0.084 vs. 0.125 for the FDG net influx rate); values in normal mice were as follows: k1=1.530, k2=3.218, k3=0.426, νB=2.8% and K_i=0.179 (Table I).

To investigate the specific molecular mechanisms underlying glucose uptake in glioma, the present study analyzed the protein expression levels of GLUT1 (Fig. 6A) in brain and tumor tissue. Compared with normal brain tissue, glioma tissue exhibited significantly elevated GLUT1 protein levels (Fig. 6B).

Morphological analysis confirmed the formation of tumors in the brain of GBM mice detected by MRI and PET-CT, along with pathological changes in tissue structure (Fig. 7A and B). HE staining showed notable heterogeneity of tumor cells (tumor cells varied in size, distinct nucleoli, abundant and eosinophilic cytoplasm, mitotic figures), accompanied by focal necrosis (Fig. 7C-E).