Distribution of glutamate and glutamine in tumor microenvironments of brain metastasis: Insights from biochemical analysis and molecular imaging

Summary

Cancer neuroscience has implicated peritumoral neurons in facilitating breast-to-brain metastasis (BrBM) progression via the N-methyl-D-aspartate receptor (NMDAR), a glutamate (Glu) receptor. The Glu-glutamine (Gln) cycle converts Glu into Gln, forming the combined Glx pool. This study investigated the spatial distribution of phosphorylated GluN2B (pGluN2B), an NMDAR subunit, and Glx in BrBM. Ex vivo analysis revealed elevated pGluN2B expression in BrBM, particularly in tumor cores, while Glx levels were paradoxically reduced. In mouse models, glutamine-based positron emission tomography (Gln-PET) imaging revealed higher tracer uptake in BrBM than that in paired breast tumors, with uptake values correlated positively with Glx concentration and pGluN2B expression. Strong Gln-PET uptake in BrBM noninvasively indicated elevated Glx metabolism in a BrBM patient, confirmed by ex vivo staining. This study highlighted the regional distribution of NMDAR and Glx, underscoring their potential as diagnostic biomarkers for BrBM. Glutamine-based molecular imaging can noninvasively visualize the tumor microenvironment relevant to cancer neuroscience.

Graphical abstract

Highlights

• •The tumor microenvironment altered the distribution of Glx and expression of NMDAR
• •Glx and its receptor NMDAR exhibit regional distribution in BrBM
• •Gln-PET noninvasively reveals the tumor microenvironment in cancer neuroscience

Molecular biology; Radiation biology; Pathology

Introduction

Brain metastases account for the majority of intracranial malignant tumors and are associated with a significantly poor prognosis.^1,2 The field of cancer neuroscience has underscored the pivotal interactions between neurons and cancer cells, which promote tumor progression.^{3,4,5,6,7,8,9,10,11,12,13,14,15} A pivotal 2019 study on breast-to-brain metastasis (BrBM) revealed that metastatic breast cancer cells upregulate the expression of the N-methyl-D-aspartate receptor (NMDAR), a glutamate (Glu) receptor composed of GluN1 and GluN2B subunits, on their cell membranes. This upregulation allows cancer cells to hijack Glu, the most abundant excitatory neurotransmitter in the brain, released by peritumoral (PT) neurons, thereby promoting tumor growth and invasion.¹⁶ Phosphorylation of GluN2B (pGluN2B) is a critical event in NMDAR signaling. Under physiological conditions, astrocytes, along with presynaptic glutamatergic neurons and postsynaptic neurons, form tripartite synapses that facilitate Glu-driven signal transmission and maintain the Glu-glutamine (Gln) cycle.¹⁷ In this dynamic cycle, Gln is converted to Glu by glutaminase (GLS) in neurons, while astrocytes utilize glutamine synthetase (GS) to convert Glu back into Gln.^{18,19,20,21,22} In BrBM, the overexpression of NMDAR hijacks Glu released by PT neurons, promoting tumor growth and invasion by forming pseudotripartite synapses and disrupting the normal Glu-Gln cycle. However, it remains unclear whether the expression of pGluN2B and concentration of Glx (Glu + Gln) vary between the tumor core (TC) and PT regions in BrBM. Uncovering these regional disparities could reveal potential diagnostic and therapeutic targets, thereby contributing to improved management of BrBM.

Positron emission tomography (PET) is a functional molecular imaging modality that visualizes biological processes with various radiolabeled tracers.²³ [18F]F-fluorodeoxyglucose ([18F]F-FDG) is among the most used tracers, identifying metabolically active regions noninvasively. However, its high baseline uptake in the normal brain limits its utility for brain tumors.²⁴ To overcome this, alternative tracers targeting amino acid metabolism, such as [18F]F-L-dihydroxyphenylalanine ([18F]F-FDOPA),²⁵ [18F]F-fluoroethyl-tyrosine ([18F]F-FET),²⁶ and [11C]C-methionine ([11C]C-MET),²⁷ or tumor-specific receptors like 68Ga-NOTA-Aca-BBN (7–14)²⁸ have been developed to provide higher specificity and tumor-to-background ratios (TBRs). Furthermore, radiolabeled Gln analogs, such as (2S,4R)-4-[18F]FGln^29,30 and (2S,4S)-4-[18F]FEBGln,³¹ can monitor Gln metabolism, indirectly reflecting pGluN2B expression. After metastasizing to the brain, tumor cells undergo significant changes in their tumor microenvironment (TME), which may influence the uptake and distribution of tracers. However, the impact of these environmental changes on tracer performance across different TMEs or tumor regions is not well understood. Investigating the use of the same tracer to target tumors in diverse TMEs and regions is crucial for understanding the dynamic interaction between the TMEs and cancer cells.³² Such insights could refine PET imaging by enhancing specificity and TBR, enabling better visualization and characterization of brain metastases. This approach has the potential to significantly improve diagnostic accuracy and inform therapeutic strategies for brain tumors.

In this study, we aimed to evaluate the potential of pGluN2B and Glx as markers for the diagnosis and treatment of BrBM. Specifically, we aimed to investigate whether the expression of pGluN2B and the concentration of Glx vary across distinct TMEs and various tumor regions, both in vitro and in vivo. By identifying these variations, we aimed to provide a foundation for understanding their potential implications for improving diagnostic accuracy and therapeutic strategies.

Results

Comparison of receptor expression and neurotransmitter concentration between breast cancer in situ and brain metastasis

Immunohistochemical (IHC) staining was employed to verify the differences in receptor expression levels between breast cancer in situ and after metastasis to the brain. Notably, the expression levels of pGluN2B (Figures 1A and 1B) were significantly elevated in the metastatic brain tumors compared to the paired in situ breast cancers (metastases: 84.95% ± 6.447% vs. in situ: 16.64% ± 6.412%, mean diff = 68.32, 95% confidence interval [CI] [62.59, 74.04], p < 0.0001, Cohen’s dz = 9.98, paired Student’s t test; Figure 1C).

Figure 1.

Comparison of pGluN2B receptor expression, glutamate, and glutamine concentration between breast cancer in situ and BrBM

(A and B) H&E staining and IHC staining of CK and pGluN2B in in situ breast cancer and metastases. Scale bars, 50 μm.

(C) Quantitative analysis of pGluN2B expression levels, as determined by IHC, in paired samples of in situ and metastatic breast cancer (t = 28.21, p < 0.0001, R² = 0.9913, n = 8 per group).

(D) The quantification of glutamate (Glu) concentration between in situ breast cancer and BrBM was analyzed using paired Student’s t test (t = 2.141, p > 0.05, ns, n = 7 per group).

(E) The quantification of glutamine (Gln) concentration between in situ breast cancer and BrBM was analyzed using a paired Student’s t test (t = 4.434, p = 0.003, R² = 0.7375, n = 8 per group).

(F) The quantification of Glx concentration between in situ breast cancer and BrBM was analyzed using a paired Student’s t test (t = 2.866, p = 0.0234, R² = 0.5434, n = 8 per group).

Furthermore, while the concentration of Glu was not found to be increased after metastasizing in the cranial (in situ: 4.432 ± 1.580 nmol/mg vs. metastases: 6.433 ± 1.544 nmol/mg, p = 0.0760; Figure 1D), the Gln concentration was significantly elevated (in situ: 0.7231 ± 0.3019 nmol/mg vs. metastases: 2.391 ± 1.172 nmol/mg, mean diff = 1.668, 95% CI [0.7783, 2.557], p = 0.003, Cohen’s dz = 1.57, paired Student’s t test; Figure 1E). Collectively, the combined pool of Glu and Gln, termed Glx, also demonstrated a significant increase in concentration in brain metastases relative to those in situ tumors (in situ: 5.184 ± 1.585 nmol/mg vs. 8.530 ± 2.678 nmol/mg, mean diff = 3.345, 95% CI [0.6048, 6.086], p = 0.0234, Cohen’s dz = 1.02, paired Student’s t test; Figure 1F).

Variation of pGluN2B expression levels and Glx from TC to normal brain tissue in BrBM

H&E staining of the whole-brain sections from mice bearing metastases revealed infiltration of tumor cells into brain tissue (Figure 2A). Based on H&E and pan-cytokeratin (CK) IHC staining (Figure 2B), the brains were segmented into three regions: TC with dense tumor cells, PT regions with infiltrated tumor cells, and normal brain tissues (NBTs). IHC staining for pGluN2B showed a decline in expression from TC to NBT (Figure 2C). The percentage of pGluN2B-positive cells in TC was significantly higher than that in PT regions (TC: 82.27% ± 8.942% vs. PT: 55.58% ± 13.81%, mean diff = 26.9, 95% CI [17.07, 36.33], adjusted p < 0.001, Cohen’s dz = 2.45) and NBT (TC: 82.27% ± 8.942% vs. NBT: 13.10% ± 7.458%, mean diff = 69.17, 95% CI [57.52, 80.83], adjusted p < 0.0001, Cohen’s dz = 5.24) (Figure 2D). Moreover, the concentrations of Glu (TC: 5.599 ± 1.063 nmol/mg vs. NBT: 6.745 ± 1.034 nmol/mg, mean diff = 1.146, 95% CI [0.0732,2.219], adjusted p = 0.0381, Cohen’s dz = 1.11, n = 8 per group), Gln (TC: 2.217 ± 1.141 nmol/mg vs. NBT: 5.069 ± 1.049 nmol/mg, mean diff = 2.852, 95% CI [1.556, 4.148], adjusted p = 0.0004, Cohen’s dz = 1.943, n = 10 per group), and Glx (TC: 8.513 ± 2.694 nmol/mg vs. NBT: 12.54 ± 2.564 nmol/mg, mean diff = 4.029, 95% CI [0.6662, 7.392], adjusted p = 0.0229, Cohen’s dz = 1.247) were consistently lower in TC than those in NBT (Figures 2E–2G). Linear regression analysis revealed a negative relationship between the percentage of pGluN2B-positive cells and Glx concentration (R² = 0.4795, p = 0.0015; Figure 2H).

Figure 2.

Variation of pGluN2B expression levels and Glx from TC to NBT in mice with BrBM

(A) H&E staining of the whole brain slide. Scale bars, 2.5 mm.

(B and C) IHC staining of CK and pGluN2B in TC, PT, and NBT. Scale bars, 50 μm.

(D) The quantification of pGluN2B in TC, PT, and NBT was analyzed using repeated-measures one-way ANOVA (F = 119.9, p < 0.0001, η² = 0.93, n = 10), followed by Tukey’s multiple comparisons test, which revealed significant differences between TC and PT (adjusted p < 0.0001), TC and NBT (adjusted p < 0.0001), and PT and NBT (adjusted p < 0.0001).

(E) The quantification of Glu concentration in TC, PT, and NBT was analyzed using repeated-measures one-way ANOVA (F = 2.702, p > 0.05, ns, n = 8 per group), followed by Tukey’s multiple comparisons test, which revealed a significant difference between TC and NBT (adjusted p = 0.0381).

(F) The quantification of Gln concentration in TC, PT, and NBT was analyzed using repeated-measures one-way ANOVA (F = 20.17, p < 0.0001, n = 10 per group), followed by Tukey’s multiple comparisons test, which revealed significant differences between TC and NBT (adjusted p = 0.0004) and TC and PT (adjusted p = 0.003).

(G) The quantification of Glu concentration in TC, PT, and NBT was analyzed using repeated-measures one-way ANOVA (F = 6.601, p = 0.0107, ns, n = 8 per group), followed by Tukey’s multiple comparisons test, which revealed a significant difference between TC and NBT (adjusted p = 0.0229).

(H) Negative correlation between Glx concentration and the percentage of pGluN2B⁺ cells.

Due to the subjective nature of surgical sampling, BrBM patient samples were segmented into TC, tumor boundary, and PT regions. The segmented samples were confirmed by H&E (Figure 3A) and CK IHC staining (Figure 3B). IHC staining for pGluN2B was also performed to verify expression variation from TC to PT regions (Figure 3C). The percentage of pGluN2B-positive cells in TC was significantly higher than that in PT (TC: 89.69% [77.95%, 97.12%] vs. PT: 3.62% [0%, 7.156%], adjusted p = 0.0050, n = 4 per group; Figure 3D). Additionally, Glu (TC: 2.299 ± 1.002 nmol/mg vs. PT: 5.207 ± 1.332 nmol/mg, mean diff = 2.908, 95% CI [0.07591, 5.7410], adjusted p = 0.0456, Cohen’s dz = 1.36), Gln (TC: 1.109 ± 0.5435 nmol/mg vs. PT: 3.453 ± 1.668 nmol/mg, mean diff = 2.344, 95% CI [0.3176, 4.371], adjusted p = 0.0295, Cohen’s dz = 1.54), and Glx (TC: 3.712 [2.817, 4.914] nmol/mg vs. PT: 6.605 [6.356, 11.75] nmol/mg, adjusted p = 0.0133) concentrations were significantly higher in PT regions than those in TC (Figures 3E–3G). Similarly, pGluN2B expression levels showed a negative correlation with Glx concentration and the percentage of positive cells in patient samples (R² = 0.6936, p = 0.0053; Figure 3H).

Figure 3.

Variation of pGluN2B expression levels and Glx from tumor core, tumor boundary, to peritumoral regions in patients with BrBM

(A) H&E staining in different regions. Scale bars, 50 μm.

(B) IHC staining of CK in different regions. Scale bars, 50 μm.

(C) IHC staining of pGluN2B in different regions. Scale bars, 50 μm.

(D) The quantification of IHC staining of pGluN2B in tumor core (TC), tumor boundary (TB), and peritumoral (PT) with Kruskal-Wallis test (H = 9.88, df = 2, p = 0.0002, η² = 0.88, n = 6), followed by Dunn’s multiple comparisons test, which showed a significant difference between TC and PT (adjusted p = 0.0050). Data were presented as mean with IQR.

(E) The quantification of Glu concentration in different regions with repeated-measure one-way ANOVA (F = 7.977, df = 2, p = 0.0165, η² = 0.6147, n = 6), followed by Tukey’s multiple comparisons test, which showed a significant difference between TC and PT.

(F) The quantification of Gln concentration in different regions with repeated-measure one-way ANOVA (F = 8.681, df = 2, p = 0.0097, η² = 0.6345, n = 6), followed by Tukey’s multiple comparisons test, which showed a significant difference between TC and PT.

(G) The quantification of Glx concentration in different regions with the Friedman test (Friedman statistic = 8.40, df = 2, p = 0.0085, n = 5), followed by Dunn’s multiple comparisons test, showed a significant difference between TC and PT.

(H) Negative correlation between Glx concentration and the percentage of pGluN2B⁺ cells.

In vivo micro-PET/CT imaging

Using mice bearing both in situ breast cancer and BrBM, the imaging efficacy of PET with (2S,4S)-4-[18F]FEBGln was investigated. The uptake of Gln-PET tracer in BrBM was significantly higher than in situ breast cancer (Figure 4A). Semiquantitative analysis of maximum standard uptake value (SUVmax) and mean standard uptake value (SUVmean) confirmed significantly higher uptake of this tracer in BrBM compared to in situ breast cancer (SUVmax: 0.2943 ± 0.06689 vs. 0.2503 ± 0.05397, mean diff = 0.044, 95% CI [0.02128, 0.06678], p = 0.0042, Cohen’s dz = 2.03; SUVmean: 0.1433 ± 0.05330 vs. 0.09557 ± 0.03357, mean diff = 0.047, 95% CI [0.02138, 0.07404], p = 0.0055, Cohen’s dz = 1.90; n = 6; Figure 4B). Both SUVmax and SUVmean showed a positive linear correlation with the Glx concentration (R² = 0.7512, p = 0.0003, Figure 4C; R² = 0.8369, p < 0.001; Figure 4D). However, SUVmax showed no correlation with pGluN2B expression levels (p = 0.0897, Figure 4E), while SUVmean exhibited a weak positive correlation with pGluN2B expression levels (R² = 0.4158, p = 0.0236, Figure 4F).

Figure 4.

In vivo micro-PET/CT imaging using (2S,4S)-4-[18F]FEBGln on breast cancer mice spontaneously with in situ and BrBM

(A) Micro-PET imaging with a BABL/c mouse bearing both metastases and in situ breast cancer using (2S,4S)-4-[18F]FEBGln. Scale bars, 1 cm.

(B) Semiquantification of SUVmax and SUVmean in in situ breast cancer and metastases (n = 6) with paired Student’s t test, which showed a significant difference in SUVs between in situ breast cancer and BrBM.

(C and D) Positive correlations between SUVs and Glx concentration. (C) R² = 0.7512, p = 0.0003; (D) R² = 0.8369, p < 0.0001.

(E and F) Correlations between SUVs and the percentage of pGluN2B⁺ cells. (E) p = 0.0897; (F) R² = 0.4158, p = 0.0236.

Strong uptake of Gln PET tracer in a BrBM patient

A 60-year-old female patient was found with an isolated focus with Gd-DTPA enhancement in the cerebellum 5 years after surgery for invasive ductal carcinoma of the breast, followed by chemotherapy, HER-2-targeted therapy, and radiotherapy (Figure 5A). She was suspected of having BrBM. No primary recurrence or metastasis to other organs except the cerebellum was found in the whole-body [18F]F-FDG PET/computed tomography (CT). Notably, the patient’s focus exhibited strong uptake of (2S,4R)-4-[18F]FGln in PET imaging (Figure 5B), with an SUVmax of 7.9. Postoperatively, H&E and IHC staining were performed on samples obtained from breast cancer and metastasis (Figures 5C and 5D). The expression level of pGluN2B in the metastasis was significantly higher than that in the primary tumor.

Figure 5.

Strong uptake of Gln PET tracer in a BrBM patient and comparison of pGluN2B expression between in situ breast cancer and BrBM

(A) Enhanced magnetic resonance imaging (MRI) of the patient suspected of BrBM.

(B) PET imaging of this patient using [18F]F-FDG and (2S,4R)-4-[18F]FGln.

(C) H&E staining and IHC staining of CK and pGluN2B in situ breast cancer obtained from the patient. Scale bars, 50 μm.

(D) H&E staining and IHC staining of CK and pGluN2B in metastases. Scale bars, 50 μm.

Discussion

Numerous studies have demonstrated that the overexpression of NMDAR in BrBM transforms PT glutamatergic neurons into accomplices in tumor growth and invasion.¹⁶ Given the presence of the Glu-Gln cycle, focusing solely on either Glu or Gln in isolation may not adequately reflect the metabolic dynamics of the TME. Therefore, we focused our analyses and discussion on Glx as a whole, rather than interpreting the two components separately. In this study, we demonstrated the overexpression of pGluN2B and elevated Glx levels in BrBM compared to primary breast cancer in animal models as a first step. Subsequently, we confirmed the regional variation of pGluN2B expression and Glx concentration using in vitro specimens from BrBM animal models and patient tissues. Finally, we employed PET imaging with radiolabeled Gln analogs, including (2S,4R)-4-[18F]FGln and (2S,4S)-4-[18F]FEBGln (Gln-PET), to assess the Glu-Gln cycle and pGluN2B expression in both animal models and patients. Notably, the results from animal models were consistent with those obtained from patients, further validating our findings. Taken together, these results reveal the potential of pGluN2B and Glx as markers for diagnosing and monitoring BrBM.

After primary breast cancer metastasizes to the brain, the overexpressed pGluN2B promotes the growth and invasion of BrBM. Our findings from in vitro specimens confirmed the heterogeneous regional distribution of pGluN2B, showing a decrease from the TC to NBT, supporting the feasibility of pGluN2B as a marker for identifying BrBM. This heterogeneity in pGluN2B expression provides foundational evidence for the subsequent design of pGluN2B-targeting molecular probes. Such probes could be valuable for preoperative diagnosis and intraoperative guidance, enhancing the precision and effectiveness of BrBM management.

Despite elevated Glx levels in BrBM, the Glx concentration in TC was lower than that observed in NBT (or PT regions in patients). Is there a contradiction in regional heterogeneity? We believe that the answer is no. The elevated Glx concentration refers to the comparison between BrBM and primary in situ breast tumors. In contrast, the regional analysis within BrBM reveals heterogeneity, with TC exhibiting relatively lower Glx than NBT. Additionally, a similar trend was reported in gliomas as well.³³ This observation showed a negative correlation with pGluN2B expression in both animal models and patient tissues. Based on previous studies,^{16,19,20,34,35} we hypothesize two probable reasons for this phenomenon: first, the overexpression of pGluN2B on BrBM cells results in the consumption of Glx; second, reduced secretion within BrBM contributes to decreased Glx production, reflecting an impaired Glu-Gln cycle. Both factors likely contribute to the observed decrease in Glx concentration from NBT to TC. This regional variation in Glx levels offers potential for utilizing Glx concentration as a diagnostic marker to identify BrBM, highlighting its clinical application potential in the diagnosis and treatment of brain metastases.

According to previous studies, Gln-PET demonstrated limited effectiveness in primary cancers such as breast cancer compared to [18F]F-FDG PET.²⁹ In mice bearing both in situ breast cancer and BrBM, Gln-PET showed significantly higher standard uptake values (SUVs) in BrBM compared to in situ tumors. Moreover, SUVmean showed a positive correlation with pGluN2B expression, confirming its link to this receptor. Consequently, Gln-PET may be a promising noninvasive tool for reflecting pGluN2B expression. Meanwhile, in a BrBM patient, Gln-PET using (2S,4R)-4-[18F]FGln exhibited superior effectiveness compared to [18F]F-FDG PET, primarily due to its limited uptake in NBTs and the resulting high TBR. Interestingly, this study innovatively investigated a PET tracer targeting cancer cells of the same origin under distinct TMEs and in different regions. Due to the presence of the blood–brain barrier, imaging and therapeutic delivery in brain tumors are generally more challenging than in extracranial tumors. Therefore, the enhanced Gln-PET performance highlights how TME changes can alter tumor biological behavior, including receptor overexpression. These findings provide direct evidence that the TME influences the uptake of molecular imaging tracers, underscoring the potential of Gln-PET in diagnosing and understanding BrBM biology and offering a foundation for TME-targeted diagnostic approaches.

This study demonstrates the differential expression of pGluN2B and the regional distribution of Glx in BrBM, highlighting the potential of Gln-PET for evaluating the TME of brain metastases. However, the current conclusions are mainly supported by existing literature and our experimental data, and the underlying mechanisms remain to be fully elucidated. In future work, we aim to explore the mechanistic basis of these observations and develop methods for real-time in vivo detection of Glx. Additionally, we plan to expand our patient cohort to further validate our findings. Interestingly, similar patterns were seen in patients with brain metastases originating from lung cancer, suggesting that the observed metabolic and receptor expression features may not be exclusive to breast cancer. This underscores the need for further investigation across brain metastases of different primary origins. Ultimately, we envision that Gln-PET could be incorporated into routine clinical workflows for the diagnosis and monitoring of brain metastases, serving as a valuable complement to conventional imaging modalities such as MRI and FDG-PET. Nevertheless, further studies are needed to clarify how Gln-PET correlates with tumor size, proliferative capacity, and growth dynamics.

In this study, the regional distributions of pGluN2B from the TC to NBT in BrBM were elucidated. Additionally, ex vivo analysis revealed a gradient in Glx concentrations from the TC to adjacent NBT. Furthermore, Gln-PET imaging was identified as a promising tool for in vivo visualization of pGluN2B expression and Glx distribution in BrBM. These findings underscore the potential of Glx and its receptor as biomarkers for BrBM, providing direct evidence that the TME influences the uptake of molecular imaging tracers.

Limitations of the study

This study has several limitations. For instance, although tissue specimens were segmented into distinct regions based on neurosurgeons’ microscopic observations and H&E staining, precise delineation remains challenging. This limitation may partly explain the lack of significant differences in neurotransmitter concentrations observed between adjacent regions. More precise methods with higher resolution should be developed to improve regional identification. Additionally, this study did not include real-time in vivo detection of Glx concentration, leaving it unclear whether the in vivo Glx distribution follows the same trend observed in vitro. Future studies should explore the real-time measurement of Glx concentration in vivo using advanced techniques such as electrochemistry or magnetic resonance spectroscopy. Moreover, the study’s sample size was limited, involving only a small number of patients. Expanding the cohort in subsequent research will be crucial for validating and generalizing the findings.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Deling Li (lideling@bjtth.org).

Materials availability

This study did not generate new reagents. All materials in this study are commercially available. Plasmids and associated vector maps generated in this study are available upon request to the lead contact.

Data and code availability

• •All data reported in this article will be shared by the lead contact upon request.
• •This article does not report original code.
• •Any additional information required to reanalyze the data reported in this article is available from the lead contact upon request.

Acknowledgments

This study received the funding support from the National Excellent Youth Science Fund Project of the National Natural Science Foundation of Scholar Program of China (grant no. 82222034) and Beijing Youth Scholar Program and Youth Talent Support Program (grant no. A002863).

Author contributions

Conceptualization, H.Z. and D.L.; methodology, J.T., Y.Z., L.C., Y.L., and H.X.; investigation, Z.W., L.Z., D.F., X.X., and L.A.; writing – original draft, J.T. and Y.Z.; writing – review & editing, H.Z. and D.L.; funding acquisition, H.Z. and D.L.; resources, H.Z. and Q.L.; supervision, Q.L., H.Z., and D.L.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Phospho-NMDAR2B (Tyr1252) Polyclonal Antibody	Thermo Fisher	Cat# 48-5200; RRID: AB_2533846
Anti-Pan cytokeratin Monoclonal Antibody	FUZHOU MAIXIN BIOTECH, LTD	Cat# Kit-0009; RRID: AB_3713159
Biological samples
BrBM from patients	N/A	N/A
Brain and in situ breast cancer samples from mice	N/A	N.A
Chemicals, peptides, and recombinant proteins
Roswell Park Memorial Institute 1640 medium	Sigma-Aldrich	R8758
10% Fetal Bovine Serum	Sigma-Aldrich	F8687
penicillin-streptomycin	Sigma-Vetec	V900929
Phosphate-Buffered Saline	Solarbio	P1020
Experimental models: Cell lines
4T1 mouse breast cancer cells	BNCC	BNCC273810
Experimental models: Organisms/strains
BABL/c mice, female	Beijing HFK bioscience company	N/A
Software and algorithms
NDP.view2	Hamamatsu Photonics K.K.	https://www.hamamatsu.com/jp/en/product/life-science-and-medical-systems/digital-slide-scanner/U12388-01.html
3D-Slicer	N/A	https://www.slicer.org
GraphPad Prism 10.5.0	Graphpad Software Inc	https://www.graphpad.com/scientific-software/prism/
Avatar 3	Pingseng Scientific	https://www.pingseng.com

Experimental model and study participant details

Animal models

Female BALB/c mice (8 weeks old, weighing 18-22g) were used to establish the animal model. Mice were purchased from Beijing HFK Biotechnology Co., Ltd., and housed under specific pathogen-free (SPF) conditions with a 12-h light/dark cycle, controlled temperature and humidity, and free access to food and water. All animal experiments were approved by the Institutional Animal Care and Use Committee of Beijing Neurosurgical Institute (BNI202401001) and conducted following the ARRIVE (Animal Research: Reporting In Vivo Experiments) guidelines. Only female mice were used due to the sex-specific nature of the breast cancer model; therefore, the influence of sex could not be assessed.

Human participants

Human tissue specimens were obtained from patients aged between 45 and 72 years undergoing surgery for brain metastases at Beijing Tiantan Hospital. Preoperative imaging confirmed intracranial lesions suggestive of breast-to-brain metastasis, and diagnoses were confirmed by intraoperative frozen section and postoperative pathology. When available, matched primary breast cancer tissues were also collected. All patients provided written informed consent. All participants were biologically female. Due to the limited number and single-sex nature of the cohort, potential sex-related differences could not be evaluated and should be explored in future studies.

Cell lines

The 4T1 murine breast cancer cell line was used for this study. Cells were cultured in RPMI-1640 medium supplemented with 10% fetal bovine serum (FBS) and maintained in a humidified incubator at 37°C with 5% CO₂. Cell lines were regularly tested for mycoplasma contamination and authenticated by short tandem repeat (STR) profiling.

Method details

Cell culture

The mouse 4T1 cell line, a well-characterized model for breast cancer research, was purchased from the Bena Culture Collection. Cells were routinely cultured in Roswell Park Memorial Institute 1640 (RPMI1640, Sigma-Aldrich) medium, supplemented with 10% Fetal Bovine Serum (FBS) and 1% penicillin-streptomycin (P/S) to ensure optimal growth conditions. Prior to experimental use, all cells were screened and confirmed to be mycoplasma-free. The cells were maintained in a humidified incubator at a temperature of 37°C with a controlled atmosphere of 5% CO2 to mimic physiological conditions and support cell viability and proliferation.

Animal studies

All animal experimental procedures were approved by the Animal Care and Use Committee of Beijing Neurosurgical Institute and strictly adhered to the ARRIVE (Animal Research: Reporting In Vivo Experiments) guidelines. Throughout the experiments, we ensured that no mouse exceeded the maximal tumor/metastasis burden as per ethical and scientific standards.

To establish an in situ breast cancer model, female Babl/c mice were inoculated with breast cancer by injecting 1x10⁵ 4T1 cells suspended in 100 μL Phosphate-Buffered Saline (PBS) into the second right mammary fat pad. Additionally, intracranial metastatic tumors were induced via a stereotactic injection of 3×10⁴ 4T1 cells in 4 μL PBS into the right hemisphere at the following coordinates: 2 mm anterior to the lambda, 2 mm lateral to the midline, and 3 mm depth from the skull hole. Mice with tumors exceeding a diameter of 1 cm were humanely euthanized to maintain ethical standards and experimental integrity.

Two weeks post-inoculation, mice underwent micro-Magnetic Resonance Imaging (MRI) to assess the formation of brain metastases. MRI scans were performed to confirm the presence and size of tumors, providing non-invasive verification of successful tumor establishment prior to further experimental procedures.

Once tumor growth was confirmed via micro-MRI, mice bearing metastases were sacrificed, and the brain was harvested for further analysis. The tumor location was visually inspected, and the brain was coronally sectioned along the tumor midline. One-half of the brain was fixed in 10% formalin for histological analysis, while the other half was processed for microscopy to distinguish the tumor core (TC), peritumoral regions (PT), and normal brain tissues (NBT), and then stored in liquid nitrogen for future molecular analysis. The in situ breast cancer tissue was also collected. After microscopic examination, tumor tissue blocks devoid of necrotic areas were selected, divided into two portions: one portion was fixed in 10% formalin, and the other was stored in liquid nitrogen.

Patients and tissue sample collection

All tissue specimens were collected from patients with a history of breast cancer who underwent surgical resection at Beijing Tiantan Hospital. Written informed consent was obtained from all participants. Preoperative MRI confirmed the presence of intracranial space-occupying lesions, and a preliminary diagnosis by the attending physician suggested a high likelihood of brain metastasis. During surgery, frozen pathology indicated a probable origin of breast cancer, which was later confirmed postoperatively as BrBM.

During resection, at least two experienced neurosurgeons identified and differentiated various regions, including the tumor core (TC), tumor boundary (TB), and peritumoral regions (PT). PT refers to the tumor-adjacent tissue located at the superficial, non-functional areas of the tumor, which must be excised during surgery. Tissues were preserved using two methods: formalin fixation for histopathological analysis and snap-freezing in liquid nitrogen for future molecular studies.

Immunohistochemistry (IHC) staining of pGluN2B

In mouse BrBM, breast cancer in situ specimens, and human tissues were fixed with 10% neutral buffered formalin and then processed into paraffin-embedded sections with a thickness of 5 μm. After deparaffinization and hydration, tissue sections were immersed in EDTA solution (pH 8.9) for antigen retrieval. Following natural cooling, sections were immersed in PBS (pH 7.4) and washed three times for 5 min on a decolorization shaker. Sections were then treated with a 3% hydrogen peroxide solution for 10 min at room temperature and washed three times with PBS. Sections were blocked with 3% BSA for 30 min at room temperature to prevent non-specific binding. After removing the blocking solution, Phospho-NMDAR2B (Tyr1252) Polyclonal Antibody (1:200, Thermo Fisher) was applied, and sections were incubated overnight at 4°C in a humidified chamber. Sections were washed three times with PBS, slightly dried, and then incubated for 2 h at room temperature with horseradish peroxidase (HRP)-conjugated anti-mouse/rabbit antibody in a humidified chamber. The DAB developing solution was added, and sections were rinsed with tap water to stop the color development. Sections were then examined under a microscope for staining, followed by H&E staining.

Quantification of glutamate (Glu) and glutamine (Gln) concentration by LC-MS/MS

A precise amount of sample was weighed using an analytical balance. Frozen samples were placed in an EP tube, and 1000 μL of extraction solution (methanol with 10% formic acid), including an isotopically labeled internal standard mixture, was added. The mixture was homogenized in an ice-water bath, followed by centrifugation at 12,000 × g for 20 min at 4°C. The supernatant (1000 μL) was transferred to a fresh tube, filtered using a 20 μm filter, and 200 μL of the filtered liquid was placed in a fresh glass vial for LC-MS analysis. The quality control (QC) sample was prepared by mixing equal aliquots of the supernatants from all samples.

LC-MS/MS analyses were performed using Agilent 1290 Infinity II LC system. Compounds were separated by HPLC on a Waters ACQUITY UPLC BEH Amide (2.1 × 100 mm, 1.7 μm, Waters). The mobile phase consisted of solvent A (1% formic acid in water, v/v) and solvent B (85% acetonitrile, v/v). The column temperature was maintained at 20 ± 1°C, and the auto-sampler temperature was 20°C. The gradient elution procedure was as follows: 0–3.5 min, 95% B; 3.5–4.0 min,80% B; 4.0–6.0 min, 62% B; 6.0–6.5 min, 55% B; 6.5–8.0 min, 95% B. The injection volume was 1 μL, and the flow rate was set at 0.3 mL/min.

The QTRAP 6500 mass spectrometry (AB Sciex) was used to acquire MS/MS spectra based on information-dependent acquisition (IDA) in LC/MS analyses. The data acquisition software (Analyst TF 1.6.2, AB Sciex) automatically collected complete scan survey MS data and triggered the acquisition of MS/MS spectra based on preselected criteria.

Probe synthesis

The radiosynthesis condition of (2S,4R)-4-[18F]FGln³⁰ and (2S,4S)-4-[18F]FEBGln³¹ was conducted following previously established methods.

Micro-PET/CT imaging

Tumor-bearing mice were injected with 150 μCi of (2S,4S)-4-[18F]FEBGln via the tail vein. Five-minute static PET scans were acquired at 90 min post-injection. Mice were anesthetized with 2% isoflurane before and during PET/CT imaging. Imaging was performed using a small-animal PET/CT scanner (Super Nova PET/CT, Pingseng Healthcare, Shanghai, China), and PET images were reconstructed using Avatar 3 (Pingseng Healthcare). In this part of the experiment, six mice were performed to repeat for comparison. Every mouse was bored both in-situ breast cancer and BrBM.

PET/CT examination of a patient with BrBM at Beijing Tiantan hospital

Following an intravenous injection of (2S,4R)-4-[18F]FGln at a dose of 3.7 MBq/kg, the patient, diagnosed with BrBM, underwent the whole-body PET/CT imaging. The scans were conducted at 60-min time point after tracer injection using a Biograph mCT Flow 64 scanner (Siemens, Erlangen, Germany) [120 kV, 146 mAs, slice: 3 mm, matrix: 200 × 200, full width at half maximum (FWHM): 5 mm, filter: Gaussian, field of view (FOV): 256 (head)]. The patient underwent a craniotomy operation for resection of the brain metastases. Postoperatively, tissue specimens were processed as formalin-fixed paraffin-embedded (FFPE). The patient had undergone total resection of the primary breast invasive ductal carcinoma five years ago, allowing for the acquisition of FFPE primary breast cancer tissue. Both the primary breast cancer and brain metastases were stained with pGluN2B.

Quantification and statistical analysis

All statistical analyses were conducted using GraphPad Prism version 10 (GraphPad Software, CA, USA). Detailed descriptions of the statistical methods, including tests used, exact sample sizes, and presentation of results, are provided in the figure legends, main text, and results sections.

The value of n is defined specifically for each type of experiment. In imaging experiments, n represents the number of biological replicates. In molecular quantification experiments, n refers to the number of samples per group. For normally distributed data, results are expressed as the mean ± standard deviation (SD); for non-normally distributed data, values are presented as the median and interquartile range (IQR). Normality of each dataset was assessed using the Shapiro-Wilk test.

For paired data with normal distribution, such as matched tumor samples from the same animal, paired two-tailed Student’s t-tests or repeated-measures one-way ANOVA were applied. When data did not meet the assumptions of normality, non-parametric tests were used: Friedman tests followed by Dunn’s multiple comparisons test for paired comparisons involving more than two groups, and Kruskal-Wallis tests followed by Dunn’s multiple comparisons test for multiple-group analyses. Homogeneity of variances was examined where necessary.

A p-value less than 0.05 was considered statistically significant. For multiple comparisons, adjusted p-values were reported accordingly. All statistical assumptions, including normality and variance homogeneity, were tested before choosing the statistical method. Where relevant, post hoc power analyses were performed to ensure sufficient statistical power for key comparisons.

Animals were randomly assigned to experimental groups using a random number generator. Blinding was implemented during data collection and analysis whenever possible. No data points were excluded unless they met pre-established exclusion criteria, such as injection failure or technical artifacts.

Published: August 14, 2025