Real-time NMR analysis of glutamine metabolism for screening anti-cancer agents in living cells
Authors: Yating Lei, Tao Huang, Xin Chai, Xinyu Fan, Sipei He, Caixiang Liu, Maili Liu, Xu Zhang
Published: 2027
ABSTRACT
Metabolic reprogramming is a substantial obstacle for anticancer drug screening, as targeted therapeutics often lose efficiency due to the dynamic adaption of cancer cells. Glutamine metabolism in cancer profoundly impacts tumor initiation, progression and metastasis. The existing agents are compromised by resistance and off-target toxicity. In this study, a real-time NMR tracking method for intracellular glutamine metabolic flux was established. This method enables comprehensive profiling of nitrogen metabolism and serves as a valuable tool for characterizing specific cancer metabolic phenotypes and screening drugs against targeted cancer cells. Applying this approach to traditional Chinese medicine (TCM) discovery, we identified Astragalus membranaceus as a potent regulator of glutamine metabolism. Through virtual screening via molecular docking, 12 potential compounds from Astragalus membranaceus were initially flagged as candidate binders toward the allosteric pocket of glutaminase 1 (GLS1). Crucially, subsequent in vitro recombinant human GLS1 enzyme activity assays successfully ruled out computational false positives and demonstrated that Compound 2 (quercetin) acts as the exclusive, direct enzymatic inhibitor among the tested monomers, capable of effectively suppressing GLS1 activity. Overall, this work provides a robust platform for real-time metabolic profiling of glutamine metabolism and drug screening at the living cell level, and offers new insights into the mechanisms of TCMs in anticancer therapy.
Introduction
Metabolic reprogramming is a hallmark feature of malignant tumors and other metabolic disorders. It refers to the adaptive changes in cellular metabolic pathways experienced by tumor cells during rapid proliferation and invasion to meet the increased demand for energy, biosynthetic precursors and redox homeostasis [1,2]. Unlike normal cells that rely mainly on oxidative phosphorylation for energy production under aerobic conditions, tumor cells often convert to glycolysis even in the presence of oxygen (Warburg effect), and exhibit abnormal activation of various anabolism pathways, including glutamine metabolism, hexosamine biosynthesis and nucleotide metabolism [3]. These metabolic changes not only support the malignant progression of tumors, but also make metabolic pathways a promising therapeutic target, because disrupting the specific metabolic dependence of tumor cells can effectively inhibit tumor growth without significantly affecting normal cells [4].
Among the various metabolic pathways involved in reprogramming, glutamine metabolism has become a key regulator, providing a key carbon and nitrogen source for the survival and proliferation of tumor cells, glutamine-mediated regulation of nitrogen flux plays an indispensable role in the physiological homeostasis and pathological progression of cells, thus becoming the focus of developing new anti-cancer strategies [5,6].
Notably, the phenomenon of "glutamine addiction" has been identified in a variety of tumor types, including gastric cancer [7] and colorectal cancer [8], with liver cancer being a prominent example. As reported by Jin et al. [9], the majority of liver cancer cell lines exhibit distinct glutamine addiction, with glutamine acting as a critical nutrient to sustain tumorigenic processes. This pronounced preference of tumor cells for glutamine is primarily driven by the elevated metabolic demands required to support their rapid uncontrolled proliferation. When it occurs to cancer, glutamine metabolism undergoes distinct reprogramming, which in turn leads to a marked upregulation of cellular glutamine demand. Furthermore, glutamine modulates nitrogen metabolism and the tricarboxylic acid (TCA) cycle, thereby enabling tumor cells to adapt to the harsh extracellular tumor microenvironment [10].
Accumulating evidence has demonstrated that the upregulation of glutamine metabolism in tumor is closely associated with anticancer drug resistance and poor clinical prognosis [11,12], which has thus drawn extensive research attention to glutamine metabolism as a promising anti-liver cancer target. Glutaminase 1 (GLS1), the rate-limiting enzyme governing glutamine hydrolysis, is highly expressed in liver cancer tissues and cells. Its abnormal overexpression is not only correlated with poor patient prognosis but also actively promotes cancer progression [13,14]. GLS1-targeted inhibitors such as CB-839 have entered phase II clinical trials, which directly verifies the great anti-cancer potential of targeting glutamine metabolism for cancer treatment [15]. Another glutamine metabolism inhibitor (GMI), V9302, which targets the glutamine transporter SLC1A5, has also become a commonly used preclinical tool for investigating glutamine metabolism in tumors [16]. Despite the promising anti-tumor activity of numerous GMIs in preclinical studies, their clinical translation and application are severely limited by the emergence of drug resistance and off-target toxicity. Therefore, developing an efficient screening strategy to identify low-toxicity and high-efficacy glutamine metabolism-targeting agents is of great significance for anti-liver cancer drug development.
A variety of analytical techniques have been widely applied in drug screening targeting tumor-specific metabolic pathways such as glucose and lipid metabolism [17,18]. Traditional analytical methodologies including enzyme-linked immunosorbent assays (ELISA), chromatography, and mass spectrometry (MS) enable high-throughput drug screening, but they lack the capacity for real-time monitoring of dynamic cellular metabolic changes. Moreover, these conventional techniques often necessitate cell lysis for sample preparation, and the subsequent extraction processes are inevitably accompanied by experimental fluctuations and interference, thereby compromising the repeatability and accuracy of experimental outcomes [19].
Metabolic reprogramming represents a typical immune escape mechanism employed by cancer cells. Consequently, methods capable of monitoring metabolic changes in living cells in real-time may offer a more effective approach for drug discovery, potentially helping to circumvent issues related to cytotoxicity and efficacy. Recent breakthroughs in the real-time monitoring of cellular metabolism have effectively addressed these inherent limitations. The advent of various small-molecule probes and fluorescent biosensors has equipped researchers with robust methodologies to monitor metabolic fluctuations within living cells [20]. Among the emerging techniques, both surface-enhanced raman scattering (SERS) and nuclear magnetic resonance (NMR) spectroscopy have been successfully employed to monitor metabolic processes in living cells [21,22]. However, SERS is limited by cellular photodamage and cannot perform real-time monitoring for more than 70 min [23], which makes it unable to capture the long-term metabolic effects of candidate drugs.
NMR spectroscopy offers a set of unique strengths for metabolic analysis. It enables non-invasive and non-destructive detection, allowing in situ monitoring of biological samples without cell disruption or chemical modification. While NMR traditionally exhibits lower sensitivity compared with MS, its detection sensitivity has been steadily enhanced in recent years through instrumental upgrades, advanced pulse sequences, and improved isotope labeling techniques [24]. Notably, the development of ultra-high-field magnets and cryogenically cooled probes has significantly improved signal-to-noise ratios. Methods such as in-cell NMR, which combines isotope labeling (e.g., 15N, 13C, 19F) with electroporation or bioreactor systems, make it possible to track dynamic metabolic changes in living cell at atomic resolution [[25], [26], [27]]. Emerging techniques including dissolution dynamic nuclear polarization (d-DNP) and 3D tissue culture systems further extend the applicability of NMR to physiologically relevant models [26,27]. Thus, NMR is powerful in screening anticancer drugs that target cellular metabolic pathways.
By employing 15N and 13C selective metabolic precursors, we can trace lots of cellular metabolic pathways using 15N or 13C NMR spectroscopy. For glutamine hydrolysis, one-dimensional version of HNCO triple resonance NMR has been used to detect glutamine metabolism [28]. However, its application has been limited to detecting only glutamine and glutamate. In this study, multiple pathways of glutamine-derived nitrogen metabolism were characterized by 1H–15N heteronuclear multiple bond correlation (HMBC) spectroscopy, enabling the detection of a broader range of nitrogen-containing metabolites [29]. The metabolic distribution and assignment of nitrogen isotopes were elucidated systematically by combining HMBC with 15N-edited 1H–13C correlation NMR (HCNCH) spectroscopy [30], thereby giving a comprehensive profiling method for glutamine-derived nitrogen metabolism. Subsequently, by coupling NMR spectroscopy with a bioreactor, we monitored the glutamine metabolic dynamics in living cells in real-time. This approach was sensitive to changes in glutamine metabolism, as validated by its effective capture of glutaminolysis alterations in HepG2 cells in the absence and presence of CB-839, a classic GLS1 inhibitor that specifically blocks glutaminolysis. This method has been used to screen efficient traditional Chinese medicines (TCMs) against liver cancer targeting glutamine metabolism. The results showed that the crude extracts of Astragalus membranaceus exerted a glutamine metabolism-modulating effect similar to that of CB-839. To unveil the specific material basis driving this enzymatic regulation, virtual screening via molecular docking was first performed, which predicted 12 candidate constituents as potential allosteric binders of GLS1. Crucially, to validate the computational hits and rigorously clear potential false positives, in vitro recombinant human GLS1 enzyme activity assays were further conducted. The biochemical results successfully filtered the candidate database and demonstrated that Compound 2 (quercetin) serves as the exclusive active monomer capable of directly and potently suppressing GLS1 enzymatic activity. Collectively, this work not only provides a novel and robust methodological approach for the real-time observation of glutamine metabolic pathways in living cells but also deepens the mechanistic insight into the anti-liver cancer effects of TCMs targeting tumor metabolic reprogramming, which delivers solid experimental evidence for the subsequent development of novel, validated glutamine metabolism-targeted anti-liver cancer leads, especially those derived from Astragalus membranaceus.
Material and methods
Experimental reagents
HepG2 cells were obtained from China Center for Type Culture Collection (CCTCC). High-glucose Dulbecco’s Modified Eagle’s Medium (DMEM) and phosphate-buffered solution (PBS) were bought from Cytiva. Fetal bovine serum (FBS), trypsin-EDTA (0.25 %), high-glucose DMEM (without glutamine) and penicillin-streptomycin solution were purchased from Gibco (Thermo Fisher Scientific, USA). Dimethyl sulfoxide (DMSO) was acquired from Sigma-Aldrich (St. Louis, USA). The stable U–13C5, U–15N2 isotope labelled glutamine (CNLM-1275, U–13C5, 99 %; U–15N2, 99 %) and D2O (deuterium purity 99.9 %) were purchased from the Cambridge Isotope Laboratories, Inc. If without stated, all labelled glutamines are carbon and nitrogen labelled. TCMs were bought from Chinese medicinal materials wholesale supermarkets (Chengdu, China) and authenticated by Professor Lifeng Han from Tianjin University of Traditional Chinese Medicine. Telaglenastat (CB-839) was purchased from MedChemExpress.
Cell culture
HepG2 cells were cultured in high-glucose DMEM supplemented with 10 % FBS and 1 % penicillin/streptomycin. They were continuously maintained in a humidified incubator with 5 % CO2 at 37 °C. Passaging was performed every 2–3 days at a 1:2 split ratio.
Crude ethanol extract of TCMs
About 100 g TCMs (Poria cocos, Sophora Flavescens, Rhizoma Atractylodes Macrocephala, Panax ginseng, Glycyrrhiza uralensis, Scutellaria baicalensis, Astragalus membranaceus, Hedyotis diffusa and Codonopsis pilosula) were soaked in 200 mL 80 % ethanol for 5 days, followed by a 15-min ultrasound extraction. The extracts were filtered through a 0.22 μm filter membrane to remove insoluble herbal residues. The extracts were concentrated by a rotary evaporator at 20 rpm and 76 °C. The concentrated solutions were frozen-dried to remove solvents, and the freeze-dried powder was stored at −20 °C. TCMs extract solutions were prepared by diluting the freeze-dried powder in DMSO sufficiently, followed by centrifuging at 12,000 rpm for 5 min.
Cell proliferation assays
In order to exclude the effect of TCMs on cell viability, Cell Counting Kit-8 (catalog number C0039; Beyotime, China) was used for testing. The performance was conducted according to the instructions. After incubating cells with Cell Counting Kit-8 for 1 h, a SpectraMax i3x plate reader was employed to measure the absorbance at 450 nm.
Extraction of HepG2 cells metabolites
HepG2 cells were cultured for 48 h, and then the old medium was discarded and replaced with fresh medium containing 4 mM isotope-labelled glutamine. After 5 h of incubation with labelled glutamine, various drugs (e.g., DMSO vehicle control) were added and the cells were incubated for another 5 h. Then the cells were harvested into a 50 mL conical tube. The cell pellet was resuspended in 600 μL PBS (containing 10 % D2O) within a 1.5 mL micro-centrifuge tube. The suspension was transferred to an ultrasonic cell disruptor and sonicated for 5 min. The whole process of ultrasonically crushing cell suspensions was conducted in an ice bath. The lysates were centrifuged at 12,000 rpm for 5 min and 500 μL of the supernatant was collected for NMR analysis.
Production of agarose threads
The production of agarose threads was implemented based on the study of Liu et al. [31] with minor modifications. In brief, the 2-hydroxyethyl agarose (Sigma-Aldrich, A4018) was added to a PBS solution at a ratio of 1.5 % W/V, repeatedly oscillated and dissolved at 90 °C. After filtration through a 0.22 μm sterile filter (Sigma-Aldrich, Millex-GP), the agarose solution was stored at 4 °C. The solidified agarose solution would be melted at 85 °C and later kept at 37 °C. One flask of 175 cm2 HepG2 cell pellet was heated at 37 °C for 20 s. The cell pellet was suspended with agarose solution and aspirated into a chromatography PEEK tube. After waiting for 2 min 20 s, the agarose threads were introduced into the NMR flow unit tube with a 5 mL syringe.
Bioreactor setup
The setting of NMR bioreactor can refer to the research of Liu et al. [31]. Briefly, there was a thermostatic unit and a circulating flow device to maintain the nutrients required by cells. 15 mL DMEM containing 10 % FBS, 1 % penicillin/streptomycin and 4 mM labelled glutamine was applied to sustain the survival of HepG2 cells and track their metabolism. Before each experiment, the bioreactor was rinsed thoroughly, ensuring the device was sterile. The rinsing process was performed in the following order: 0.2 M sodium hydroxide, 3 M citric acid, 0.2 M sodium hydroxide, 75 % ethanol, and sterile water, each at the same flow rate of 0.5 mL/min for 10 min.
NMR spectroscopy
All NMR spectra were acquired at 310 K on a 600 MHz Bruker Avance III spectrometer equipped with a 5 mm TCI (13C, 1H, 15N) CryoProbe. For the 1H–15N HMBC experiment [29], spectral widths were set to 12 ppm and 185 ppm for the 1H and 15N dimensions, with carrier frequencies centered at 4.7 ppm and 90 ppm, respectively. Data were collected with 48 transients using 2048 and 128 time-domain points for the 1H and 15N dimensions, respectively. The 1H–15N polarization transfer delay was optimized to 100 ms, corresponding to 1/(2J) for a long-range coupling constant of 5 Hz, to maximize signal intensity. For the HCNCH experiment [30], spectral widths were 12 ppm (1H) and 185 ppm (13C), with carrier frequencies centered at 4.7 ppm and 90 ppm, respectively. Data were collected with 64 transients using 2048 and 128 time-domain points for the 1H and 13C dimensions, respectively. The polarization transfer delays were optimized based on an H–C coupling constant of 140 Hz, while the 13C–15N transfer delay was specifically set to 25 ms, corresponding to 1/(4J) for a C–N coupling constant of 10 Hz, to ensure optimal signal transfer. The spectral width of 13C dimension was narrowed in real-time NMR experiments of screening drugs to improve spectral resolution, whose signals assignment can refer to 1H dimension.
Molecular docking
Ligand-flexible docking was performed using AutoDock Vina to evaluate the binding affinities and interaction modes. The main components of Astragalus membranaceus and their three-dimensional structures were retrieved from the Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP, https://www.tcmsp-e.com/) [32]. The crystal structure of human GLS1 (PDB ID: 5HL1) was downloaded from the RCSB Protein Data Bank (https://www.rcsb.org/). Crucially, to accurately represent the physiological state and the allosteric binding pocket located at the dimer-dimer interface, the tetrameric biological assembly of GLS1 was utilized. Prior to docking, protein and ligand preparations were conducted using AutoDock Tools (ADT) 4.2.6. For the receptor, the original co-crystallized ligand (CB-839) and water molecules were removed, and polar hydrogens and Gasteiger charges were then added to prepare the receptor. For the ligands, hydrogens and charges were computed, and their rotatable bonds and torsion angles were defined using the ADT Torsion Tree utility. To specifically target the allosteric binding site, the docking grid box was set to 28.9 × 19.1 × 15.7 Å, with grid center positioned at X = −17.6, Y = −42.9, and Z = 2.0. The exhaustiveness parameter was set to 8, and a maximum of 20 binding modes were generated for each ligand.
GLS1 enzyme activity assay
GLS1 activity was determined using a commercial assay kit obtained from Solarbio (Catalog, BC1455). Compound 1, 2, 3, 5, 6, 7, 9, 10 and 12 were purchased from MedChemExpress. Briefly, GLS1, His-tag Recombinant (BPS Bioscience, Catalog, 71102) were diluted in GLS extraction buffer, followed by a 20-min incubation with different drugs. The enzymatic activity was calculated based on the formula supplied by the kit.
Results and discussion
Metabolic profiling of glutamine metabolites
The exchange broadening caused by proton change in water, the low natural abundance of 15N and the lack of 15N NMR reference limit the application of NMR in nitrogen metabolism [33]. 1H–15N HMBC and HCNCH were selected to address these challenges. 1H–15N HMBC was used to detect the correlation between 15N nucleus and remote 1H nucleus in nitrogen-containing compounds [29]. This method escaped the direct detection of protons attached to nitrogen, thus avoiding signal losses due to proton exchange. The HCNCH method was also employed, which realizes coherent transfer of 1H → 13C → 15N through INEPT magnetization transfers modulated by strong heteronuclear J coupling [30]. Consequently, it can sensitively provide 1H–13C chemical shift correlation information for nitrogen-containing metabolites, facilitating their identification [30]. Compared with previous studies [28,29], the combination of these two methods allowed a more comprehensive analysis of glutamine metabolic flux.
To obtain a comprehensive profile of glutamine metabolites, cells were cultured in fresh DMEM medium supplemented with 4 mM U–13C5, U–15N2 isotope labelled glutamine for 24 h. These cells were then thoroughly lysed by ultrasonication, and the supernatant was subsequently analyzed by NMR spectroscopy with the addition of 10 % D2O for field locking.
As shown in Fig. 1, 19 and 13 signals were observed in the HMBC and HCNCH spectra, respectively. These metabolites were assigned by combining HCNCH analysis of the same sample (Fig. 1b.) with relevant literature references [29,33,34]. Detailed information is compiled in Table 1. It was observed that glutamine provided both carbon and nitrogen for aspartate, asparagine, proline, uracil, uridine and UDP-N-acetyl-hexosamine (UDP-HexNAc) while providing only nitrogen for alanine and some unknown metabolites in HMBC. The unknown metabolites were predicted to be purine or pyrimidine derivatives, which are essential for understanding potential glutamine metabolic pathways. These identified metabolites were consistent with the glutamine metabolic pathways described in the literature [35,36], indicating multiple pathways in which glutamine may be involved. Specifically, the production of α-ketoglutarate, aspartate, asparagine and proline from glutamine is catalyzed by glutaminase and glutamate dehydrogenase and various transferase [[36], [37], [38], [39]], demonstrating active glutamine catabolism for energy production and biosynthetic precursor generation. Alanine was synthesized from glutamine via alanine transaminase, indicating pyruvate availability and active transamination. Additionally, the synthesis of uracil, uridine and UDP-HexNAc was identified, showing complete channeling of glutamine nitrogen into nucleotide and glycosylation precursors, which indicates high demand for nucleic acid synthesis and protein glycosylation in cancer cells [40,41].Fig. 1NMR spectra of cell lysate. (a) 1H–15N HMBC spectrum. (b) HCNCH spectrum.Table 1Metabolite assignments from NMR spectroscopy for cell lysate and living cell.Table 1No.metabolitesassignmentsδH (ppm)δC (ppm)δN (ppm)1AlanineCH31.488//αNH2//43.02AsparagineβCH22.87337.4/βCH22.94937.4/αNH2//40.1-CONH2//111.53AspartateCH22.691//CH22.811//CH3.90554.9/αNH2//39.64GlutamateβCH22.053//γCH22.38233.5/αNH2//40.75GlutamineβCH22.143//γCH22.45733.5/CH3.76557.3/αNH2//40.8-CONH2//112.26ProlineC(4)H23.42348.8/C(4)H23.34448.8/C(1)H4.13363.8///55.67PyroGluγCH22.416//(pyroglutamate)//130.58UracilN–CH CH5.814//N–CHCH7.535146.2/N–CHCH//134.2OC–NH–CO//159.29UridineC–CH of uracil5.914//N–CH of uracil7.870144.4///145.710UDP-HexNAcNA-H2.02624.6/(Uridine diphosphate -N-acetyl-hexosamine)NA-H2.06124.6/N–CH of uracil7.947144.2/11U11.408////123.912U22.032////131.813U32.066////123.314U45.664////134.115U55.958////134.116U67.729////131.217U77.934////131.918U88.102////123.6U1–U8, unknown metabolites;/, undetected signals.
NMR spectra of cell lysate. (a) 1H–15N HMBC spectrum. (b) HCNCH spectrum.
Table 1Metabolite assignments from NMR spectroscopy for cell lysate and living cell.Table 1No.metabolitesassignmentsδH (ppm)δC (ppm)δN (ppm)1AlanineCH31.488//αNH2//43.02AsparagineβCH22.87337.4/βCH22.94937.4/αNH2//40.1-CONH2//111.53AspartateCH22.691//CH22.811//CH3.90554.9/αNH2//39.64GlutamateβCH22.053//γCH22.38233.5/αNH2//40.75GlutamineβCH22.143//γCH22.45733.5/CH3.76557.3/αNH2//40.8-CONH2//112.26ProlineC(4)H23.42348.8/C(4)H23.34448.8/C(1)H4.13363.8///55.67PyroGluγCH22.416//(pyroglutamate)//130.58UracilN–CH CH5.814//N–CHCH7.535146.2/N–CHCH//134.2OC–NH–CO//159.29UridineC–CH of uracil5.914//N–CH of uracil7.870144.4///145.710UDP-HexNAcNA-H2.02624.6/(Uridine diphosphate -N-acetyl-hexosamine)NA-H2.06124.6/N–CH of uracil7.947144.2/11U11.408////123.912U22.032////131.813U32.066////123.314U45.664////134.115U55.958////134.116U67.729////131.217U77.934////131.918U88.102////123.6U1–U8, unknown metabolites;/, undetected signals.
Metabolite assignments from NMR spectroscopy for cell lysate and living cell.
U1–U8, unknown metabolites;/, undetected signals.
The observed metabolic profile is highly consistent with cancer cell characteristics. The broad incorporation of 15N from glutamine into amino acids, nucleotides, and nucleotide sugars mirrors the metabolic dependency observed in cancer cells, as tumors usually exhibit glutamine addiction, where glutamine becomes an essential nutrient for survival and proliferation. Furthermore, the upregulation of pyrimidine synthesis in rapidly proliferating cells to meet nucleotide demands for DNA replication provides additional evidence of active glutamine-driven pyrimidine biosynthesis in cancer cells [41]. The observation of 15N-labelled UDP-HexNAc indicates high demand for glycosylation precursors, consistent with an aggressive cancer phenotype.
In a word, multiple pathways of glutamine metabolism indicate the hybrid metabolic state typical of cancer cells. As cancer cells are usually characterized by distinct metabolic personalities, this NMR-based comprehensive profiling of nitrogen metabolism may enable characterization of specific cancer cell types and provide active or multi-pathway targets for anticancer drug development.
Real-time monitoring of intracellular glutamine metabolites
To monitor dynamic changes of intracellular glutamine metabolites in real time, a bioreactor system was used to circulate 4 mM U–13C5, U–15N2 isotope labelled glutamine through an NMR tube containing HepG2 cells fixed in gel. Real-time NMR detection was achieved continuously during medium circulation. Due to its higher magnetization transfer efficiency compared to 1H–15N HMBC, the HCNCH experiment was utilized, and the results are shown in Fig. 2. Although the sensitivity of real-time detection was limited, as evidenced by the disappearance of many signals observed in cell lysate, the clear detection of UDP-HexNAc, glutamine, glutamate, pyroGlu indicates that the experiment captured information from multiple glutamine metabolic pathways.Fig. 2The dynamic variation of metabolite signals over time in living cells observed by HCNCH spectrum.
The dynamic variation of metabolite signals over time in living cells observed by HCNCH spectrum.
The substrate glutamine remained nearly constant due to the high concentration and volume of circulated glutamine in the medium. In contrast, all other detected nitrogen metabolites showed time-dependent enhancement over the 10 h perfusion period (the acquisition duration of each HCNCH spectrum is 2 h), indicating active glutamine metabolism and accumulation of downstream products. These results suggest that the developed method enables real-time monitoring of multiple nitrogen metabolic pathways in living cells, and may serve as a valuable tool for investigating cell line-specific metabolic properties and their responses to different environmental conditions.
The viability of the cells was verified using 31P NMR spectroscopy. The corresponding spectra of ATP levels acquired before and after NMR detection were almost identical (Fig. S1), suggesting that cells maintained a stable metabolic state throughout the real-time detection period.
Screening of traditional Chinese medicine targeting glutamine metabolism in living cells
To evaluate whether this method can serve as a valuable tool for investigating nitrogen metabolic responses to different environmental conditions. We employed it to detect dynamic changes in nitrogen metabolic pathway of HepG2 cells under three different conditions: (1) control, (2) in the presence of CB-839, a sensitive inhibitor of glutaminolysis 1 (GLS1) that targets a key enzyme in cancer cell glutaminolysis, and (3) in the presence of some traditional Chinese medicines (TCMs) suggested to possess nitrogen modulating properties. All inhibitors and TCMs were added 4 h after cells loading into the magnet. In control experiments, DMSO was used as a vehicle control in place of inhibitor or TCMs.
As expected, rapid glutamine consumption and glutamate production were observed in the control group (Fig. 3a and b). In contrast, CB-839 significantly suppressed glutamate accumulation, with levels decreased dramatically following CB-839 addition at 4 h (Fig. 3a and b). This confirms the effectiveness of the inhibitor [42,43], and suggests that this method can successfully detect glutamine metabolism inhibition.Fig. 3Metabolic changes in living cells with different agents over time. Metabolic changes on glutamine levels (a) and glutamate levels (b) after treatment with DMSO, CB-839 (5 μM) and crude extracts of TCMs (1 mg/mL). Effects recurrence of DMSO, CB-839 (5 μM) and crude extracts of Astragalus membranaceus (1 mg/mL) on glutamine levels (c) and glutamate levels (d), each group was repeated three times. Influences at 8 h of DMSO and 0.5, 1, 2.5 mg/mL crude extracts of Astragalus membranaceus on glutamine levels (e) and glutamate levels (f), each group was repeated three times. Statistical significance was assessed by one-way ANOVA compared with DMSO. ∗: P-value <0.05.
Metabolic changes in living cells with different agents over time. Metabolic changes on glutamine levels (a) and glutamate levels (b) after treatment with DMSO, CB-839 (5 μM) and crude extracts of TCMs (1 mg/mL). Effects recurrence of DMSO, CB-839 (5 μM) and crude extracts of Astragalus membranaceus (1 mg/mL) on glutamine levels (c) and glutamate levels (d), each group was repeated three times. Influences at 8 h of DMSO and 0.5, 1, 2.5 mg/mL crude extracts of Astragalus membranaceus on glutamine levels (e) and glutamate levels (f), each group was repeated three times. Statistical significance was assessed by one-way ANOVA compared with DMSO. ∗: P-value <0.05.
Upon the addition of crude extracts of three different TCMs, Scutellaria baicalensis, Codonopsis pilosula, and Astragalus membranaceus, the levels of glutamine decreased similarly across all treatments, but the production of glutamate varied markedly (Fig. 3a and b). Glutamate gradually increased with treatment of Scutellaria baicalensis or Codonopsis pilosula, mirroring the control, suggesting that this TCM extract dose does not possess any inhibitory effect against glutamate. Conversely, the increasing trend of glutamate was arrested with Astragalus membranaceus and CB-839 after the addition of them, indicating that they can effectively inhibit the level of glutamate. This suggests that Astragalus membranaceus contains inhibitors against glutamate production. To obtain a reliable estimate of the inhibitors’ potency, these effects were validated through three independent biological replicates. (Fig. 3c and d), confirming that both CB-839 and Astragalus membranaceus inhibit glutamate production.
Dose gradients of Astragalus membranaceus extract were set at 0 mg/mL (DMSO vehicle control), 0.5 mg/mL, 1 mg/mL, and 2.5 mg/mL. The relative levels of glutamine and glutamate at 8 h were shown in Fig. 3e and f, respectively. Compared with the DMSO group, glutamine levels remained unchanged across all treatment conditions, indicating that glutamine availability was sufficient to meet cellular metabolic demands. Glutamine levels in the 0.5 mg/mL and 1 mg/mL treatment groups were similar to each other and lower than in the DMSO group (Fig. 3e). The 2.5 mg/mL treatment resulted in a significantly greater reduction in glutamate levels compared to DMSO, indicating a gradual decrease in the levels of glutamate with an increasing concentration of Astragalus membranaceus crude extraction (Fig. 3f)
Importantly, all TCMs showed low cytotoxicity under these experimental conditions (Fig. S2). Our results indicate that Astragalus membranaceus is a promising candidate with low toxicity for cancer through modulation of glutamine metabolism, with GLS1 as a potential molecular target.
Identification and virtual screening of potential active components in the screened Chinese herbal medicine via molecular docking
Astragalus membranaceus and its bioactive components possess various pharmacological properties, including immunomodulation, hepatoprotection, and anti-tumor activities [44]. To investigate the chemical entities within Astragalus membranaceus that potentially contribute to its regulatory effects on glutamine metabolism via GLS1, we screened the major chemical constituents from the TCMSP database and performed molecular docking studies aimed at the well-established allosteric pocket of GLS1 (the binding site of CB-839).
Generally, a lower (more negative) binding energy indicates a stronger binding affinity between the ligand and the target protein. In this study, twelve promising small-molecule ligands were initially selected as candidate binders based on the following stringent criteria: docking score ≤ −7.5 kcal/mol, oral bioavailability (OB) ≥ 30 %, and drug-likeness (DL) ≥ 0.18 [45]. Detailed structural and docking profiles of these candidates are summarized in Table 2.Table 2The structure and docking affinity of CB839 and active molecules from Astragalus membranaceus.Table 2No.TCMSP ID/Chemical NameChemical StructureScaffold/SubclassScore (kcal/mol)ControlCB-839−12.71MOL000354/isorhamnetinFlavonoids/flavonols−9.22MOL000098/quercetinFlavonoids/flavonols−9.03MOL000422/kaempferolFlavonoids/flavonols−9.04MOL000378/7-O-methylisomucronulatolFlavonoids/isoflavanones−8.85MOL000417/calycosinFlavonoids/flavones−8.56MOL000392/formononetinFlavonoids/flavones−8.37MOL000380/(6aR,11aR)-9,10-dimethoxy-6a,11a-dihydro-6H-benzofurano[3,2-c]chromen-3-olFlavonoids/pterocarpins−7.98MOL000398/isoflavanoneFlavonoids/isoflavanones−7.99MOL000438/(3R)-3-(2-hydroxy-3,4-dimethoxyphenyl)chroman-7-olFlavonoids/isoflavanones−7.910MOL000296/sitosterolSteroids/sterols−7.611MOL000374/5′-hydroxyiso-muronulatol-2′,5′-di-O-glucosideFlavonoids/isoflavanones−7.612MOL000239/jaranolFlavonoids/flavonols−7.5
Table 2The structure and docking affinity of CB839 and active molecules from Astragalus membranaceus.Table 2No.TCMSP ID/Chemical NameChemical StructureScaffold/SubclassScore (kcal/mol)ControlCB-839−12.71MOL000354/isorhamnetinFlavonoids/flavonols−9.22MOL000098/quercetinFlavonoids/flavonols−9.03MOL000422/kaempferolFlavonoids/flavonols−9.04MOL000378/7-O-methylisomucronulatolFlavonoids/isoflavanones−8.85MOL000417/calycosinFlavonoids/flavones−8.56MOL000392/formononetinFlavonoids/flavones−8.37MOL000380/(6aR,11aR)-9,10-dimethoxy-6a,11a-dihydro-6H-benzofurano[3,2-c]chromen-3-olFlavonoids/pterocarpins−7.98MOL000398/isoflavanoneFlavonoids/isoflavanones−7.99MOL000438/(3R)-3-(2-hydroxy-3,4-dimethoxyphenyl)chroman-7-olFlavonoids/isoflavanones−7.910MOL000296/sitosterolSteroids/sterols−7.611MOL000374/5′-hydroxyiso-muronulatol-2′,5′-di-O-glucosideFlavonoids/isoflavanones−7.612MOL000239/jaranolFlavonoids/flavonols−7.5
The structure and docking affinity of CB839 and active molecules from Astragalus membranaceus.
While the highly optimized control inhibitor CB-839 exhibited the strongest binding affinity (−12.7 kcal/mol), the 12 screened natural compounds also demonstrated robust binding to GLS1, with scores ranging from −9.2 to −7.5 kcal/mol. Furthermore, structural analysis of these top candidates revealed a striking scaffold convergence. Remarkably, 11 out of the 12 candidate compounds belong to the flavonoid family. Based on their specific chemical structures (Table 2), these flavonoids can be further classified into four distinct subclasses: flavonols (Compounds 1, 2, 3, and 12), isoflavanones (Compounds 4, 8, 9, and 11), flavones (Compounds 5 and 6), and pterocarpins (Compound 7). This rich diversity of flavonoid derivatives initially provided a structural database of candidate scaffolds for allosteric pocket fitting [46]. Based on the computational models, these flavonoid scaffolds were predicted to align with the critical flexible loop (residues 316–326) and the adjacent helix at the dimer-dimer interface, theoretically forming hydrogen bonds with loop residues (such as Phe322 and Lys320) and establishing hydrophobic interactions with Tyr394.
It is worth noting that while molecular docking indicated that these 12 constituents from 4 structural scaffolds could fit into the allosteric pocket of GLS1, virtual screening inevitably carries a risk of false positives due to the simplified solvent models and static or semi-flexible protein structures used in computation. Therefore, direct biochemical validation is mandatory to eliminate computational artifacts and pinpoint the true active ingredients.
In vitro enzymatic validation of candidates and identification of compound 2 as the core GLS1 inhibitor
In this study, molecular docking served as an efficient computational filter to screen potential active compounds derived from Astragalus membranaceus for potential GLS1-targeting activity. To experimentally validate these computational hits and pinpoint the genuine principal active components, the direct inhibitory effects of 9 commercially available candidate monomers were evaluated via in vitro recombinant human GLS1 enzyme activity assays (Fig. 4). Strikingly, the biochemical assay provided a sharp refinement and filtering of the virtual screening results, demonstrating an exquisite selectivity: among all the tested candidates, only Compound 2 (quercetin) exhibited powerful and direct inhibition against GLS1 activity, whereas the other 8 tested compounds showed negligible inhibitory effects under identical conditions. The enzymatic inhibitory capacity of Compound 2 at the tested concentration was highly pronounced. This single-hit outcome suggests that while multiple flavonoid matrices stereochemically fit the allosteric pocket during semi-flexible computer simulations, only the specific substituent decoration and functional group alignment unique to Compound 2 can successfully induce the critical conformational entrapment required to shut down GLS1 enzymatic catalysis, thereby efficiently clearing the computational false positives.Fig. 4Inhibitory effects of Astragalus membranaceus compounds (100 μM) and CB-839 (25 μM) on GLS1 enzyme activity. DMSO was used as a negative control. Statistical significance was assessed by one-way ANOVA compared with DMSO. ∗: P-value <0.05, ∗∗: P-value <0.01.
Inhibitory effects of Astragalus membranaceus compounds (100 μM) and CB-839 (25 μM) on GLS1 enzyme activity. DMSO was used as a negative control. Statistical significance was assessed by one-way ANOVA compared with DMSO. ∗: P-value <0.05, ∗∗: P-value <0.01.
Compound 2 has been previously identified from Astragalus membranaceus by UPLC-MS/MS, and flavonoids in roots are known to be highly abundant [47]. Also, Compound 2 has a variety of physiological activities, including antioxidant, anti-inflammatory, anti-cancer, antiviral and antibacterial effects [48]. Its powerful effect in regulating GSH levels has been verified by Myhrstad et al. [49]; however, its relationship with glutamine metabolic pathway, especially GLS1 enzyme, has not been reported.
This finding validates that Astragalus membranaceus naturally contains bioactive substances capable of directly targeting and regulating the GLS1 enzyme. As a key metabolic regulatory enzyme, GLS1 participates in critical cellular glutamine metabolism processes, and its abnormal activation is closely linked to the occurrence and progression of multiple diseases. The identified Compound 2 serves as a promising lead compound targeting GLS1. This study rigorously confirms the direct enzymatic suppression and targeted binding affinity of Compound 2 against GLS1, clearing computational artifacts and laying a solid foundation for subsequent structural optimization and natural GLS1 inhibitor development.
Conclusions
Although many drug candidates demonstrate efficacy in vitro, they often fail in clinical practice, potentially due to the inability to capture continuous metabolic response and off-target effects in the cellular environment. Real-time NMR spectroscopy addresses this limitation by enabling long-term and continuous evaluation of candidate potency in living cells. In the present study, we systematically elucidated the nitrogen metabolic distribution and glutamine metabolism characteristics in HepG2 cells. Multiple glutamine metabolic pathways were detected, indicating this method provides comprehensive profiling of nitrogen metabolism and serves as a valuable tool for characterizing specific cancer metabolic phenotypes and screening drugs against targeted cancer cells.
As a proof-of-concept application, Astragalus membranaceus was identified as a potent inhibitor of glutamine metabolism, as evidenced by significant suppression of glutamate production. To unravel the direct molecular principal components behind this effect, virtual screening via molecular docking was initially used to predict candidate flavonoid scaffolds targeted at the allosteric pocket of GLS1. Crucially, subsequent cell-free enzymatic assays successfully filtered the computational hits, effectively eliminating false positives and confirming that Compound 2 (quercetin) acts as the exclusive constituent capable of directly and potently suppressing GLS1 enzymatic activity.
In conclusion, our findings provide a comprehensive metabolic profile of glutamine metabolism and establish a robust platform for drug screening and metabolic pathway regulation at the living cell level. While this study establishes a foundation for metabolic intervention by Astragalus membranaceus, further validation across diverse cancer cell models and dose-dependent efficacy and safety analyses are warranted. Future work should focus on pre-clinical efficacy evaluation to translate these findings into therapeutic applications.
CRediT authorship contribution statement
Yating Lei: Writing – review & editing, Writing – original draft, Visualization, Validation. Tao Huang: Writing – review & editing, Data curation. Xin Chai: Writing – review & editing, Investigation, Data curation. Xinyu Fan: Writing – review & editing, Software. Sipei He: Writing – review & editing, Data curation. Caixiang Liu: Writing – review & editing, Writing – original draft, Validation, Methodology. Maili Liu: Writing – review & editing, Supervision, Funding acquisition. Xu Zhang: Writing – review & editing, Visualization, Funding acquisition, Conceptualization.
Data statement
Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact Caixiang Liu (cxliu@apm.ac.cn).
Declaration of competing interest
The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Maili Liu is an editorial board member of Magnetic Resonance Letters but was not involved in the editorial review or the decision to publish this article.