Photorespiration is linked to DNA methylation by formate as a one-carbon source

nature plants Article https://doi.org/10.1038/s41477-026-02222-x Photorespiration is linked to DNA methylation by formate as a one-carbon source Received: 22 May 2025 Accepted: 8 January 2026 Valentin Hankofer 1,2,7, Andrea Ghirardo3, Lisa Obermaier4, Gernot Poschet 5, Jisha Suresh Kumar1, Inonge Gross1, Jörg Durner1, Michael Rychlik4, Markus Wirtz 5, Rüdiger Hell 5, Jörg-Peter Schnitzler 3 & Martin Groth 1,6,8 Published online: xx xx xxxx Check for updates Photorespiration is a costly cellular process that reduces photosynthetic efficiency. While mitigating photorespiratory losses could boost crop yields, the interconnection of photorespiration with other processes is increasingly recognized. Its high carbon turnover generates mitochondrial one-carbon (C1) metabolites, including formate, but their contribution to cellular C1 metabolism has remained unclear. DNA methylation is an important epigenetic modification that depends on methyl groups provided by folate-mediated C1 metabolism. Here we show that photorespiration supplies C1 units for DNA methylation in Arabidopsis. We demonstrate that carbon from formate is incorporated into 5-methylcytosine through the C1-tetrahydrofolate synthase pathway, which operates predominantly during the day. Elevated CO2 that suppresses photorespiration alters the methylome, especially when the serine-derived C1 supply, which compensates for a blocked formate-derived supply, is compromised. These findings establish a metabolic link between photorespiration and epigenome stability and provide a framework for understanding methylome dynamics under rising CO2 levels and other environmental influences on photorespiration. Plants undergo photorespiration because ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco) poorly discriminates between CO2 and O2 (ref. 1). Changes in CO2 levels, temperature and light intensity alter carboxylation and oxygenation rates, making photorespiration highly dynamic2. Detoxification of the oxygenation product and carbon salvage through the photorespiratory pathway consume ATP and release CO2, thereby reducing net photosynthesis by up to 50% (ref. 3). Reducing oxygenation and improving photorespiratory efficiency can therefore increase crop yields3. However, photorespiration is tightly connected to other cellular processes, including folate-mediated one-carbon metabolism (FOCM)2,4,5. During photorespiration, glycine accumulates and is transported into mitochondria, where it is oxidized by the glycine decarboxylase complex (GDC), transferring a C1 unit to tetrahydrofolate (THF)6,7 (Fig. 1a). In parallel, mitochondrial serine hydroxymethyltransferase (SHMT) converts glycine to serine, regenerating THF8,9. A portion of the C1 units generated by the GDC supports mitochondrial FOCM, while surplus C1 is released by 10-formyl-THF deformylase (FDF) as formate to sustain flux and avoid the build-up of photorespiratory intermediates10. Mitochondrial formate dehydrogenase readily oxidizes accumulating formate to CO2 (ref. 11). While activated C1 units are not exchanged between compartments, both serine and formate produced in mitochondria can supply cytosolic C1 metabolism: serine via cytosolic SHMT, generating 5,10-methylene-THF, and formate via 10-formyl-THF synthetase (THFS)4,8,9,12 (Fig. 1a). 5,10-methylene-THF directly serves Institute of Biochemical Plant Pathology, Helmholtz Munich, Neuherberg, Germany. 2TUM School of Life Sciences, Technische Universität München, Freising-Weihenstephan, Germany. 3Research Unit Environmental Simulation, Helmholtz Munich, Neuherberg, Germany. 4Lehrstuhl für Analytische Lebensmittelchemie, TUM School of Life Sciences, Technische Universität München, Freising-Weihenstephan, Germany. 5Centre for Organismal Studies Heidelberg, Heidelberg University, Heidelberg, Germany. 6Institute of Functional Epigenetics, Helmholtz Munich, Neuherberg, Germany. 7Present address: Leibniz Institute of Vegetable and Ornamental Crops, Großbeeren, Germany. 8Present address: Biomedical Center, Molecular Biology Division, Ludwig-Maximilians-Universität München, Planegg-Martinsried, Germany. e-mail: 1 Nature Plants Article https://doi.org/10.1038/s41477-026-02222-x 1a 3 5-Formyl-THF 7b 10-Formyl-THF Serine Methionine 4 1a 10 WT LD 9 Formate 5 4 3 2 1 0 8 10 12 14 16 18 1.0 0 mCG ratio in DMRs A4 0.5 0 1.0 A2 40 20 b b mt t WT 0 m 24% 32% 29% 23% 40% A3 b 22% 39% HV gbM RdDM CMT2 Other m Rest A5 6% 1.0 A2 21% 5% 5% m mt t WT –1 log2(TPM + 1) 5 10 15 mCG ratio 12% 49% 0 6% 4% 0.5 A5 A6 A7 1 0 0 0.5 0 m mt t WT 60 2% 5% 7% 1% 0 A4 A3 11% SDCpro–GFP e Classes 26% 0.5 1.0 A1 Cluster A1 a –1 kb 1.5 1.0 0.5 0 −0.5 −1.0 −1.5 80 Correlation mCG z score 10-Formyl-THF THF Days after germination d 7b Glycine c Leaf area under LD (cm2) t CO2 + NH3 Serine 8 b mt Glycine Formate THF m 5,10-Methenyl-THF 7b 1b 5,10-Methenyl-THF TTS 5-Methyl-THF Hcy Glycine 5 SAM 5,10-Methylene-THF 2 6 7a 5,10-Methylene-THF 1 kb DNA methylation SAH Mitochondria Thymidylate synthesis TSS Cytosol/nucleus CTCF × 104 a 0.5 1.0 A6 A7 33% m mt t WT m mt t WT Fig. 1 | Growth defects and loss of transcriptional silencing in mthfd1-1 are suppressed by thfs. a, Cross-compartmental biochemical pathways of FOCM. (1a) and (1b) SHMT; (2) MTHFR; (3) methionine synthase; (4) S-adenosyl methionine (SAM) synthetase; (5) SAM-dependent methyltransferase, including DNA methyltransferases; (6) SAH hydrolase (SAHH); (7a) methylenetetra hydrofolate dehydrogenase and (7b) methenyltetrahydrofolate cyclohydrolase (MTHFD); (8) THFS; (9) 10-FDF; (10) GDC. Hcy, homocysteine. b, Representative pictures of 3-week-old mthfd1-1 (m), mthfd1-1 thfs double mutant (mt), thfs (t) and wild-type (WT) plants grown under LD conditions, and leaf area quantification from automated phenotyping (right). Scale bars, 1 cm. The data are presented as mean values ± s.d. (n = 8). c, Corrected total cell fluorescence (CTCF) from SDCpro-GFP expression quantified via confocal laser scanning microscopy. The box plots represent the median (centre lines), the 25th (bottom) and 75th (top) percentiles, and the minimum and maximum points within 1.5× the interquartile range (IQR) (n = 10). Lowercase letters represent significant differences (P < 0.05, one-way analysis of variance followed by post-hoc Tukey test). See Supplementary Table 8 for the P values. d, Hierarchically clustered heat map of mean scaled mCG ratios (z scores) of all DMRs (rows) from pairwise comparisons to the WT (left); distributions of mCG ratios in four main clusters, A1 (n = 9,959), A2 (n = 2,537), A3 (n = 245) and A4 (n = 268) (centre); and per cent overlaps of DMRs with hypervariable DNA methylation (HV) sites, CMT2-dependent methylated sites, RNA-directed DNA methylation (RdDM) sites, gbM sites and the remaining reference mthfd1-1 hypo-DMRs (Other m), as well as the remaining non-overlapping DMRs (Rest) (...truncated)