Mouse studies

Animal experiments were either subject to review by the University of Ghent Animal Ethics Committee or were performed in accordance with UK Home Office Regulations (project licenses 70/8645 60/4181, PP6345023 and PP0604995) and subject to review by the Animal Welfare and Ethical Review Board of the University of Glasgow. Mice were housed in ventilated cages with ad libitum food and water access and 12-h light/12-h dark cycles. The temperature of the facility was kept between 19 and 23 °C with 55 ± 10% humidity. Glultm3Whla fl mice8 in a C57BL/6 mixed background were kindly provided by M. Mazzone (KU Leuven). Glultm3Whla fl mice were bred in-house and were backcrossed (N5–N12) into a C57BL/6J background, and all other mice were on a mixed C57BL/6J background unless otherwise stated. No statistical calculation was done to determine the sample size. Comparable numbers of male and female Glultm3Whla fl mice received a single injection of AAV8-TBG-Cre virus (2 × 1011 genomic copies per mouse) at ~60 d of age, and samples were collected at ~120 d of age unless indicated otherwise. Mice heterozygous for Ctnnb1lox(ex3) (ref. 28) received a single injection of AAV8-TBG-Cre virus or AAV8-Null-Cre (2 × 1011 genomic copies per mouse) at 61–71 d of age. After 4 d, the mice were culled, and tissue and blood were sampled. Male and female mice carrying the Ctnnb1lox(ex3)/wt and Rosa26DM.lsl-MYC/DM.lsl-MYC,29 alleles without or with Glultm3Whla fl alleles received a single injection of AAV8-TBG-Cre virus (6.4 × 108 genomic copies per mouse) at ~80 d of age. Triple mutants were backcrossed into C57BL/6J N10. Male mice backcrossed into C57BL/6J N10 carrying the Rosa26DM.lsl-MYC/DM.lsl-MYC and Trp53tm1brn/tm1brn alleles received a single injection of AAV8-TBG-Cre virus (5 × 1010 genomic copies per mouse) at ~57 d of age32. Mice were randomly assigned to experimental groups based on their genotypes, and, with the exception of MRI quantification, the analyses were not performed blindly.

Germ-free mice

Axenic 8-week-old C57BL6/J mice from the Ghent Germ-Free and Gnotobiotic Mouse Facility at Ghent University were transferred from flexible film isolators (NKP) to positive-pressure isocages (Tecniplast). Colon and cecum content from one 8-week-old C57BL6/J SPF mouse was isolated under anaerobic conditions and homogenized in 5 ml of sterile PBS with 0.1% l-cysteine. The suspension was left for 5 min to let particulates settle, and the supernatant was transferred to a 50-ml Falcon tube and used as donor material. Five germ-free C57BL6/J mice received an oral gavage with 200 µl of SPF donor microbiota suspension, and six germ-free C57BL6/J mice received an oral gavage with sterile PBS with 0.1% l-cysteine. Both groups were housed in positive-pressure isocages for 3 weeks before mice were killed and tissue was collected.

Stable isotope tracing, methylamine administration and GS inhibition in vivo

Wt/wt and Δ/Δ mice were injected intraperitoneally (IP) with 2 g per kg (body weight) U-13C6 glucose (CLM-1396, Cambridge Isotopes) or 2 mmol per kg (body weight) 13C-methylamine (277630, Sigma-Aldrich), and tissue samples were collected 30 min and 2 h after injections, respectively. Wt/wt mice were IP injected with 200 mg per kg (body weight) U-13C5-glutamine (CLM-1822, Cambridge Isotopes). Shortly before injection, blood was collected by tail vain puncture and immediately diluted 1:50 in the extraction solution for LC–MS analysis. Thirty minutes after injection, mice were killed, and blood was collected by cardiac puncture and processed as described above. Methylamine (426466, Sigma-Aldrich) was supplemented in the drinking water (0.1% (wt/vol)) to 6- to 9-week-old wt/wt and Δ/Δ mice for 5 months. Water consumption was not affected by methylamine supplementation, and age-matched mice not administered methylamine were used as controls.

Six- to 9-week-old wt/wt mice were injected IP either with NaCl 0.9% (vehicle solution) or 10 mg per kg (body weight) MSO (M5379, Sigma-Aldrich). Blood was collected by tail vain puncture shortly before (0 h) and 2, 4, 8 and 24 h after MSO injection and immediately diluted 1:50 in the extraction solution for LC–MS analysis. Tissue samples were collected for analysis 4 h after MSO administration.

MALDI imaging

Serial sections of wt/wt and Δ/Δ livers were cut at 10-μm thickness, processed for standard immunohistochemistry (IHC) for GS and OAT or mounted on IntelliSlides (1868957, Bruker) for MALDI imaging. Sections from wt/wt and Δ/Δ livers were paired on each slide. Freeze-dried sections were shipped to the Bruker Daltonics facilities and sprayed with N-(1-naphthyl)-ethylenediamine dihydrochloride matrix with a TM-sprayer (HTX Technologies). Data were acquired on a timsTOF fleX instrument (Bruker) in negative Q-TOF ion mode at a 10-µm pixel size. Metabolic compounds were automatically annotated using MetaboScape 2021b (Bruker), and ion distributions were visualized with SCiLS Lab 2021c (Bruker).

Magnetic resonance imaging and urine collection

At 121–156 d after AAV8-TBG-Cre administration, urine samples were collected from Cnntb1lox(ex3)/wtRosa26DM.lsl-MYC/DM.lsl-MYC mice and immediately diluted 1:50 in LC–MS extraction solution. Within 24 h of urine collection, MRI images were acquired on a nanoScan PET/MRI (Mediso) using the 35-mm radiofrequency coil. A non-gated T1-weighted gradient echo sequence in the coronal/sagittal plane was acquired in three dimensions with a repetition time of 20 ms, echo time of 3.8 ms and flip angle of 15°. The image matrix is non-isotropic with dimensions of 179 × 512 × 60 and a field of view of 3.58 × 10.24 × 3.00 cm. Standard Fourier transform was used for reconstruction. No postprocessing was performed, and manual segmentation was performed blindly using VivoQuant ver4.0 (Invicro).

Cell lines

HEK293 and HepG2 cell lines were obtained from ATCC. T16 cells were kindly provided by S. Niclou (Luxembourg Institute of Health). HEK293 and HepG2 cells were routinely cultured with MEM with 1 g liter–1 glucose (21090022, Gibco) supplemented with 2 mM glutamine (Gibco), 1% non-essential amino acids (11140035, Gibco), 1 mM pyruvate (S8636, Sigma-Aldrich) and 10% FBS (10270106, Gibco). T16 cells were routinely cultured in MEM (21090022, Gibco) supplemented with 1% non-essential amino acids (11140035, Gibco), 0.65 mM glutamine (A2916801, Gibco), 0.1 mM pyruvate (S8636, Sigma-Aldrich), 1% ITS (1 g liter–1 insulin, 550 mg liter–1 transferrin, 670 μg liter–1 sodium selenite; 41400045, Gibco), 10 ng ml–1 epidermal growth factor (EGF; AF-100-15, Peprotech), 10 ng ml–1 fibroblast growth factor (FGF; AF-100-18B, Peprotech), 6.8 µg ml–1 vitamin B12 (V6629, Sigma-Aldrich), 2 µg ml–1 heparin (H3393, Sigma-Aldrich) and 400 mg liter–1 AlbuMAX II (11021029, Thermo Fisher Scientific). For the experiments with HEK293 and HepG2 cell lines, cells were seeded in complete MEM, and 15–24 h later, the medium was replaced with Plasmax33 supplemented with 2.5% FBS (10270106, Gibco) and 200 mg liter–1 AlbuMAX II (11021029, Thermo Fisher Scientific) for an additional 24 h before the experiments. For the experiments, T16 cells were cultured in Plasmax33 supplemented with AlbuMAX II (11020029, Thermo Fisher Scientific), ITS, FGF, EGF and heparin at the concentrations used for routine culture. All cell lines were authenticated using the Promega GenePrint 10 kit (B9510, Promega) and tested negative for mycoplasma infection using the MycoAlert mycoplasma detection kit (LT07-318, Lonza).

CRISPR–Cas9 and overexpression constructs

Non-targeting control sequence and two guide RNAs (gRNAs) against exon 3 of GLUL were cloned into LentiCRISPRv2 (refs. 34,35) using BsmBI restriction enzyme. For each sequence, 3 × 105 HepG2 or HEK293 cells were transfected with 1 µg of gRNA using jetPrime (Polyplus). Twenty-four hours after transfection, the media were replaced and supplemented with 1 µg ml–1 puromycin for selection. Puromycin-resistant cells (500) were seeded in a 25-cm cell culture dish. Individual clones were collected, and GS expression was tested by immunoblotting. The following gRNA target sequences were used: non-targeting control gRNA, 5′-GTAGCGAACGTGTCCGGCGT-3′; GLUL gRNA 1, 5′-TCTGTAGGTCCATATTACTG-3′; GLUL gRNA 2, 5′-TTCTAGTGGGAATTTCAGAT-3′. The plasmids containing wt, R324A and R324C GLUL cDNA were kindly provided by U. Kutay (ETH Zurich)36. The GLUL coding sequence was subcloned from pcDNA5/FRT/TO/HA into pcDNA3.1 NEO (+) using 5′-BamHI and 3′-EcoRI restriction sites. However, these vectors lack the Kozak sequence upstream of the start codon of GLUL, which is required for efficient translation of the full-length GS protein. To add the wt Kozak sequence to the GLUL expression cassette, the plasmids were linearized by BamHI and BsrGI digestion, and a repair oligonucleotide was ligated into the plasmid. These repair oligonucleotides contain the sequence for five nucleotides upstream of the GLUL start codon (5′-CCACCATG-3′). The sense sequence was

5′-ATCCACCATGACCACCTCAGCAAGTTCCCACTTAAATAAAGGCATCAAGCAGGT-3′, and the antisense sequence was

5′-GTACACCTGCTTGATGCCTTTATTTAAGTGGGAACTTGCTGAGGTGGTCATGGTG-3′. GLUL sequences were confirmed by Sanger sequencing.

Enzymatic assays

Enzymatic assays were performed using a modified protocol described previously37,38. Briefly, 2 µg of human recombinant GS obtained from Novus Biologicals (NBP2-52376) or purified GS obtained as described in the next section was added to the reaction mixture consisting of 100 mM imidazole/HCl (pH 7.2; I5513, Sigma-Aldrich), 50 mM glutamate (pH 7.2; G5889, Sigma-Aldrich), 20 mM ATP (pH 7.2; A7699, Sigma-Aldrich) and 20 mM MgCl2 (M4880, Sigma-Aldrich). Unless otherwise indicated, 500 µl of reaction mixture was incubated for 2 min at 37 °C, and the reaction was initiated by adding NH4Cl (A9434, Sigma-Aldrich), CH3NH2·HCl (M0505, Sigma-Aldrich), 15NH4Cl (299251, Sigma-Aldrich) or 13C-methylamine (277630, Sigma-Aldrich) at the concentrations indicated in Fig. 4f,g or in the legends of Extended Data Fig. 3a,b. Aliquots (5 µl) of the reaction mixtures were sampled at the times specified in the figure legends, diluted 1:1,000 in LC–MS extraction solution and analyzed by LC–MS. Km values were calculated in GraphPad 9.4 (Prism) by using a Michaelis–Menten equation. N2-Methyl-l-glutamine used in Fig. 2j was synthesized by replacing glutamate with 40 mM N-methyl-l-glutamate (ICN15555583, Fisher Scientific) in the reaction mixture described above. The reaction was incubated at 37 °C for 2 h, and an aliquot of the mixture was diluted 1:1,000 for LC–MS/MS analysis.

Expression and purification of human wt GS and R324C mutant

Full-length human GLUL wt and R324C mutant were amplified using 5′ primer (5′-TAAGCAGGATCCACCACCTCAGCAAGTTCCCACTTAAATAAAGGC-3′) with BamHI and 3′ primer (5′- TGCTTAGAATTCTTAATTTTTGTACTGGAAGGGCTCATCGCCGG-3′) with EcoRI from pCDNA3.1 GS–HA. Following double digestion and agarose gel purification, GLUL fragments were ligated into pRSF-DUET containing 12×His-SUMO, and the sequence was confirmed with Sanger sequencing. Chemically competent Escherichia coli BL21(DE3) Rosetta2 pLysS (Novagen) cells were transformed with 12×His-SUMO–GS in pRSF-DUET. Cell cultures were grown in Luria–Bertani medium supplemented with 1 mM MgSO4 at 37 °C to an optical density at 600 nm of ~0.8 and induced with 0.35 mM isopropyl β-d-1-thiogalactopyranoside; expression occurred overnight at 16 °C. Cells were collected, centrifuged (600g), resuspended in IMAC buffer A (25 mM sodium phosphate, 500 mM NaCl and 50 mM imidazole, pH 7.5) and lysed with a microfluidizer at 10,000 psi. The lysate was cleared by spinning at 19,000 r.p.m. for 45 min at 4 °C, syringe filtered using a 0.45-μm filter and loaded onto a 5-ml His-Trap HP column (GE Life Sciences). The loaded column was washed for 30 column volumes (cv) in IMAC buffer A and eluted with 100% IMAC buffer B (25 mM sodium phosphate, 500 mM NaCl and 350 mM imidazole, pH 7.5) for 5 cv. Fractions were combined and dialyzed against ULP1 cleavage buffer (25 mM Tris, 500 mM NaCl and 5 mM β-mercaptoethanol, pH 8.0) overnight at 4 °C in 3,500 molecular weight cutoff (MWCO) SnakeSkin dialysis tubing (Thermo Fisher) with 5 μM SUMO protease (ULP1). The cleavage reaction was filtered using a 0.45-μm syringe filter and pass backed over a 15-ml His-Trap HP column (GE Life Sciences). The flowthrough was concentrated in a 10,000-MWCO Amicon centrifugal filter unit (Merck Millipore) to a final volume of 3 ml and loaded on a 26/600 Superdex 200 SEC column (GE Life Sciences) preequilibrated in SEC buffer (20 mM HEPES, 300 mM NaCl and 0.5 mM TCEP, pH 7.5). Full-length GS eluted ~0.59 cv, and pure fractions confirmed by SDS–PAGE were concentrated to 67–117 μM using a 10,000-MWCO Amicon centrifugal filter unit. Protein aliquots were stored at −80 °C until use.

LC–MS metabolomics

Cells were extracted as previously described33. Briefly, cells plated in six wells were washed three times with ice-cold PBS and incubated for 5 min at 4 °C with 400 µl of LC–MS extraction solution (20% water, 50% methanol and 30% acetonitrile).

Tissue fragments (20–40 mg) were extracted by using the Precellys Evolution homogenizer (Bertin) and 25 µl of LC–MS extraction solution per mg of tissue. Cell and tissue extracts were centrifuged at 16,000g for 10 min at 4 °C, and the supernatant was stored at −74 °C until LC–MS analysis. Compound peak areas obtained for cells were normalized on the total micrograms of proteins determined for each extracted well with a modified Lowry assay33.

Metabolites from the biological extracts were injected (5 μl) and separated using a ZIC-pHILIC column (SeQuant; 150 mm × 2.1 mm, 5 μm; Merck) coupled with a ZIC-pHILIC guard column (SeQuant; 20 mm × 2.1 mm) using an Ultimate 3000 HPLC system (Thermo Fisher Scientific). Chromatographic separation was performed using a 15-min linear gradient starting with 20% ammonium carbonate (20 mM, pH 9.2) and 80% acetonitrile, terminating at 20% acetonitrile at a constant flow rate of 200 μl min–1. The column temperature was held at 45 °C.

A Q Exactive Orbitrap mass spectrometer (Thermo Fisher Scientific) equipped with electrospray ionization was coupled to the HPLC system for both metabolite profiling and metabolite identification. For profiling, the polarity switching mode was used with a resolution (RES) of 35,000 or 70,000 at 200 m/z to enable both positive and negative ions to be detected across a mass range of 75 to 1,000 m/z (automatic gain control (AGC) target of 1 × 106 and maximal injection time (IT) of 250 ms).

Data-dependent fragmentation was performed to aid metabolite identification using a wt/wt liver pooled sample comprised of a mixture of all sample extracts analyzed per experimental batch. The Q Exactive was operated in positive and negative polarity mode separately (35,000 RES, AGC target of 1 × 106 and max IT of 100 ms), and the ten most abundant ions were chosen for fragmentation (minimum AGC target of 1 × 103, AGC target of 1 × 105, max IT of 100 ms, 17,500 RES, stepped normalized collision energy of 25, 60 and 95, isolation width of 1 m/z, dynamic exclusion of 15 s and charge exclusion of >2) per survey scan.

Data-independent fragmentation was performed to acquire fragmentation spectra of specific metabolites including 5-methylglutamine (positive polarity, m/z 161.0920). Fragmentation spectra were continuously recorded with the following parameters: 17,500 RES, isolation width of 0.7 m/z, AGC target of 1 × 105, max IT of 250 ms and stepped normalized collision energy of 25, 60 and 95.

Untargeted metabolomics analysis was performed using Compound Discoverer software (Thermo Scientific v3.2). Retention times were aligned across all data files (maximum shift of 2 min and mass tolerance of 5 ppm). Unknown compound detection (minimum peak intensity of 1 × 106) and grouping of compounds were performed across all samples (mass tolerance of 5 ppm and retention time tolerance of 0.7 min). Missing values were filled using the software’s ‘Fill Gap’ feature (mass tolerance of 5 ppm and signal/noise tolerance of 1.5). Compound identification was assigned by matching the mass and retention time of observed peaks to an in-house library generated using metabolite standards (mass tolerance of 5 ppm and retention time tolerance of 0.5 min) or by matching fragmentation spectra to mzCloud (www.mzcloud.org; precursor and fragment mass tolerance of 10 ppm and match factor threshold of 60).

Targeted metabolomics analysis was performed using Tracefinderv4.1 (Thermo Scientific), and the peak areas of metabolites were determined by using the m/z of the singly charged ions (extracted ion chromatogram, ±5 ppm) and the retention time from our in-house metabolite library.

N5-Methylglutamine was quantified in the serum and urine samples by a standard addition method. The concentrations of d,l-N5-methylglutamine indicated in Extended Data Fig. 4h, m–o were obtained by spiking a stock solution of the compound to a solution extracted and pooled from the respective fluid samples (n = 4 mice for serum and n = 5 mice for urine). The same method was used to quantify N5-methylglutamine in wt/wt liver samples (n = 1) spiked with d,l-N5-methylglutamine to obtain final concentrations of 0, 1, 5, 10 and 20 µM. To estimate the micromolar concentration of N5-methylglutamine in liver tissue, we used a ratio of 1 mg of wet tissue per µl. The samples used for the quantification of serum concentrations were analyzed with the Q Exactive operated in positive-selective ion monitoring mode (70,000 RES, AGC target of 2 × 105, max IT of 240 ms and m/z of 161.0919 ± isolation window of 1 m/z) using the same chromatographic conditions as above. All other samples were analyzed as described above for biological extracts.

Methylamine was quantified with an LC–MS method adapted from previous reports39,40. An aliquot of 25 µl of mouse serum was transferred to an Eppendorf tube, and 5 µl of trichloroacetic acid (20% in water) was added and mixed by vortexing for 30 s. The samples were centrifuged at 12,000g for 10 min, and 15 µl of the supernatant was transferred to a new tube and supplemented with 22.5 µl of borate buffer (0.5 M, pH 11) and 12.5 µl of tosyl chloride (10 mg ml–1 in acetonitrile). The mixture was mixed by vortexing for 5 s and incubated for 2 h at 50 °C. The samples were cooled down at room temperature and analyzed by LC–MS. A selected reaction monitoring mode was used to detect derivatized methylamine on an Altis QQQ mass spectrometer equipped with a Vanquish LC system (Thermo Fisher Scientific). Chromatography was performed on an Acquity HSS T3 column (Waters; 150 mm × 2.1 mm, 1.8 μm). The mobile phase consisted of solvent A (water with 0.1% formic acid) and solvent B (acetonitrile with 0.1% formic acid). Separation of metabolites was performed with the following gradient: 0 min 20% B, 8 min 95% B and 10 min 20% B at a constant flow rate of 0.3 ml min–1. The injection volume was 5 µl. Three transitions were optimized using a standard of derivatized methylamine from the positive precursor ion (m/z 185.9) to product ions (m/z 64.8, 90.8 and m/z 154.9). The total cycle time was 0.8 s, and Q1 RES was (full-width at half-maximum) 0.7 and Q3 RES was (full-width at half-maximum) 1.2. For each transition, the collision energy applied was optimized to generate the greatest possible signal intensity and using the calibrated RF values. The optimized source parameters were a spray voltage of 3,500 V, sheath gas of 35, aux gas of 7, ion transfer tube temperature of 325 °C and vaporizer temperature of 275 °C. Data acquisition was performed using Xcalibur 4.1 (Thermo Scientific) software, and quantification was performed using Tracefinderv4.1 (Thermo Scientific).

Ammonia measurement

Ammonia concentration was measured in frozen sera or in blood collected from the tail vein of mice and immediately analyzed with the blood ammonia meter PocketChem BA PA-4140 (Arkray).

Immunohistochemistry

Tissue samples were fixed in a 10% solution of neutral buffered formalin overnight (16–24 h) and transferred to 70% ethanol. Paraffin-embedded tissue blocks were cut into 5-μm sections and stained with hematoxylin and eosin or with the following antibodies: GS (1:800; HPA007316, Sigma-Aldrich) and OAT (1:200; ab137679, Abcam). IHC images were visualized with an Aperio ImageScope v12.4 (Leica Biosystems).

Immunoblotting

Cells were washed twice with ice-cold PBS, and proteins were extracted with RIPA buffer (20-188, EMD Millipore) containing protease and phosphatase inhibitors (A32961, Thermo Fisher Scientific). Protein amounts were quantified with a standard bicinchoninic acid assay (A32961, Pierce). Tissues were extracted with 25 µl of RIPA buffer per mg of wet weight. Tissue fragments were homogenized with the Precellys Evolution homogenizer (Bertin), and 20–80 μg of protein extract was loaded in 9.5% acrylamide gels for electrophoresis and blotted onto nitrocellulose membranes. PageRuler prestained protein ladder (26616, Thermo Fisher Scientific) was used as a reference for the protein molecular weight. Membranes were incubated overnight with the following primary antibodies: GS (1:1,000; 610517, BD Bioscience), OAT (1:1,000; ab137679, Abcam), β-actin (1:1,000; ab8229, Abcam), β-tubulin (1:2,000; T5201, Sigma-Aldrich) and vinculin (1:1,000; V9131, Sigma-Aldrich). The secondary antibodies anti-rabbit horseradish peroxidase (1:1,000; 7074, Cell Signaling Technology), anti-mouse IRDye 800CW (1:2,500; 926-32212, Licor) and anti-goat IRDye 680CW (1:2,500; 926-68074, Licor) were used, and membranes were imaged with an Odyssey infrared scanner and visualized with Image Studio Lite 5.2 (Licor) or imaged with Clarity Western ECL substrate (1705061, Bio-Rad) and a Chemidoc MP imager (Bio-Rad) and visualized with Chemidoc Image Lab 6.0 (Bio-Rad). Scanned images of uncropped membranes are shown in the Source Data.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.