In Hangzhou, five sixths of the kidney tissue was cut out of male mice five or six weeks old. Twelve weeks later the hearts of those mice weighed more than the hearts of sham-operated mice. The left ventricular wall had thickened and the heart muscle cells were wider. The group of Junnan Wu, at Sir Run Run Shaw Hospital of Zhejiang University, looked inside those cells.
The cardiomyocyte is the cell that makes the heart contract, packed with mitochondria and hungry for energy. In the hearts of mice with kidney disease the cardiomyocyte had changed its fuel. Glutaminolysis was switched on, and glutaminolysis is the pathway that splits glutamine into glutamate to feed the Krebs cycle. Glutamine in the heart fell and glutamate rose, week after week. Oxidative phosphorylation, the way a mitochondrion draws almost all of its energy, was the most suppressed pathway of all.
The enzyme that cuts glutamine is called glutaminase 1, GLS1. GLS1 was more abundant and more active already six weeks after surgery, before the wall was clearly thickened. Fibrosis of the myocardium arrived only at eighteen weeks. In a different model of an enlarged heart, the aorta tied off with a silk suture, GLS1 did not move.
The team then looked for what switches GLS1 on, and the candidate came out of patients' blood. Among urea, uric acid, parathyroid hormone and five other substances that build up when kidneys stop filtering, phosphate raised GLS1 activity more than any other. Under high phosphate the cardiomyocytes made more ammonia and held less ATP, with mitochondria losing membrane potential. Giving a pyruvate analogue, which helps clear ammonia, put some of this back.
GLS1 was deleted from cardiomyocytes alone, by a genetic cross, and the hypertrophy shrank. CB-839, a GLS1 inhibitor, was given by mouth at two hundred milligrams per kilogram twice a day from week twelve to week eighteen. Here too the heart grew less, with less mitochondrial damage. Every in vivo experiment was done in male mice, and the title of the paper says so. The authors write that they can conclude nothing about females. An inhibitor tested in male mice at that dose is not a therapy for people, and the paper does not claim it is.
The human part is a separate matter and should be kept separate. At five Chinese hospitals, between 17 March and 7 December 2022, 308 peritoneal dialysis patients were recruited. Of these, 228 reached the analysis, with a median follow-up of 2.52 years. Serum taken at entry was put on neonatal rat cardiomyocytes, and GLS1 activity was measured on those cells. The outcome was the left ventricular mass index, the mass of the left ventricle estimated from the echocardiogram and divided by body surface area.
In the third of patients with the highest GLS1 activity the index grew by a mean of 16.5 grams per square metre. In the lowest third it fell by 5.7 and in the middle third by 8.2, with standard deviations between 37 and 62. The correlation between GLS1 activity and change in the index is 0.216, and a value like that accounts for about 5 per cent of the variability. The authors say it plainly, this is a surrogate assay on rat cells and not GLS1 activity in the patients' own hearts.
The paper announces a composite outcome in which death before the second echocardiogram counts as the worst result. Deaths do not appear in the recruitment diagram, and all 228 patients have a change in the index. Another number fails to add up. The interventricular septum of those patients measures a mean of between 17.3 and 18.5 millimetres across the three bands. The posterior wall of the same hearts sits between 11.6 and 12.0 millimetres, and the hypertrophy threshold the authors use is twelve millimetres of posterior wall.
The phosphate poured on the cells deserves a last line. In the forty patients of the proteomic substudy, blood phosphorus reached 2.4 millimoles per litre. The cells were treated with 2.5 millimoles per litre, and the dose-response curve runs up to 5. Of the 862 serum proteins, 66 came out associated with GLS1 activity before correction for multiple testing, and by chance alone 43 were expected. After the correction not one is left.
The heart under study was male, and the phosphate in the dishes was higher than the phosphate in the patients' blood.

In the cohort of 228 peritoneal dialysis patients, split into three bands of GLS1 activity measured on serum taken at entry, the left ventricular mass index grew by a mean of 16.5 grams per square metre in the top band and fell in the other two. The dot is the mean, the bar is one standard deviation and not a confidence interval, and the bars overlap heavily. This is the human, observational part of the work, while the mechanism was studied in male mice. Chart produced by ScienceOnline from data in Bai L. et al., Nature Communications 2026, doi 10.1038/s41467-026-79129-w, open access under CC BY-NC-ND 4.0. Original newsroom chart.
References
Bai L., Wang Y., Zhou L., et al., Mao J., Wu J., «Metabolic reprogramming modulates cardiomyocyte hypertrophy via GLS1 in chronic kidney disease in male mice», Nature Communications, 2026, Article in Press, received 9 November 2025 and accepted 12 October 2026, online 1 November 2026, open access under a Creative Commons CC BY-NC-ND 4.0 licence. doi: 10.1038/s41467-026-79129-w
Raw data deposited by the authors, RNA-sequencing in GEO under accession GSE343701, serum proteomics in PRIDE under PXD077955, metabolomics and imaging in OMIX under OMIX019621, reanalysed public dataset GSE180852. Individual patient data from the cohort are not deposited and are available on request to the corresponding author.



