Cell biology textbooks may need revising: mitochondria do not simply release ATP into the cytoplasm to diffuse like Wi-Fi. A 2026 Nature paper (DOI: 10.1038/s41586-026-10588-3) by Hesham A. Sadek's team (University of Arizona, UT Southwestern, CNIC Madrid, NUS, and others) shows that mitochondria physically dock onto the nuclear pore complex and deliver energy through a dedicated line.
Key points
- The plug and socket: VDAC1 (voltage-dependent anion channel 1) on the mitochondrial outer membrane binds RANBP2 (Nup358), a component of the nuclear pore's cytoplasmic filaments. The interaction was independently discovered by GST pulldown (using RANBP2's C-terminal domain as bait) and BioID proximity labeling (90 specifically labeled proteins, VDAC1 among them).
- Structural validation: AlphaFold predicts a VDAC1 loop inserting into a groove of RANBP2-CTD like a plug into a socket. Point mutations (RANBP2 FD→AA; VDAC1 ETT→LII) weaken or abolish the interaction and reduce mitochondria–nucleus contact sites.
- Energy is shipped as phosphocreatine: When the connection is disrupted, nuclear ATP and phosphocreatine (PCr) drop while cytoplasmic and mitochondrial ATP remain normal — the power plant works, but the transmission line is cut. The working model: mitochondrial ATP is converted to PCr near the pore, PCr travels through the VDAC1–RANBP2 channel, and nuclear creatine kinase regenerates ATP for phosphorylation, DNA replication, transcription, and chromatin remodeling.
- Downstream collapse in knockouts: Phosphoproteomics of RANBP2-knockout cells shows widespread loss of nuclear protein phosphorylation (histone modification, differentiation, transcriptional regulation). RNA-seq, ChIP-seq (H3K27me3), and ATAC-seq jointly show reduced chromatin accessibility, increased repressive H3K27me3, and downregulation of cardiac development, differentiation, and neuronal development gene programs.
- Lethal in vivo: CRISPR truncation of RANBP2's C-terminal domain in mice causes embryonic death at E10.5, with apoptosis in the heart region, reduced cardiomyocyte cross-sectional area, and neural crest differentiation defects. Rapidly dividing, differentiating tissues depend most on the dedicated line; passive ATP diffusion may suffice in low-metabolic cells.
- Half a billion years of conservation: Across 182 vertebrate and 164 arthropod species, RANBP2's C-terminal domain is nearly invariant (Shannon entropy near zero in vertebrates), indicating that evolution strongly preserves this energy-coupling hub. RANBP2's STRING network links nuclear transport (Nup88, Nup214, CRM1), cell cycle (SUMOylation, topoisomerase IIα), and mitochondria (VDAC1, Cox11, hexokinase I).
- RNA-seq: GSE325290
- ChIP-seq: GSE324951
- ATAC-seq: GSE324952
- Proteomics: PXD065792, PXD065793
- Biophysical model: https://colab.research.google.com/drive/10ufpBhsLk96DidzzFEsj79KGXr2iGR64
- Evolution analysis: https://github.com/akwestfall/EvolutionRANBP2
Why it matters
The paradigm shift: organelle energy transfer is active, specific, and physically wired — joining mitochondria–ER (MAMs), mitochondria–lipid droplet, and mitochondria–lysosome contacts as a structural connection, but one that is essential for embryonic development. Potential implications include heart failure (cardiomyocyte energy deficit), neurodegeneration, a new angle on the Warburg effect (cancer cells losing the mitochondria–nucleus line), and regenerative medicine (boosting the connection could accelerate iPSC differentiation). More broadly, cell fate may depend not only on transcription factors but on organelle wiring and energy allocation — chromatin cannot open without a local ATP supply.
Evidence quality
The paper stacks eight layers of evidence: proteomics (GST pulldown + BioID), structure (AlphaFold + mutagenesis), super-resolution imaging (confocal + STED), phosphoproteomics, metabolomics (HPLC + MS), RNA-seq, ChIP-seq/ATAC-seq, and a CRISPR mouse model. All data and code are public: