Ferroptosis is an iron-dependent, non-apoptotic form of cell death defined by the catastrophic accumulation of lipid peroxides in cellular membranes. Unlike apoptosis, which cancer cells frequently evade through well-characterized mutations, ferroptosis depends on the lipid composition of the cell, and one enzyme sits at the heart of that dependency: acyl-CoA synthetase long-chain family member 4, commonly abbreviated ACSL4. ACSL4 shapes the membrane pool of oxidizable polyunsaturated fatty acids, and cells with high ACSL4 levels are markedly more vulnerable to ferroptotic death. The new work demonstrates that pancreatic cancer cells actively destroy ACSL4 under hypoxic conditions, and that this destruction is orchestrated by an unexpected player drawn from the machinery of valine, an essential branched-chain amino acid.

The researchers began by comparing tumor samples from patients who responded to gemcitabine-based neoadjuvant chemotherapy with those from non-responders. Metabolomic and transcriptomic profiling revealed that valine, leucine, and isoleucine degradation was significantly enriched in non-responders, and within that pathway the gene encoding MMSDH, ALDH6A1, stood out for its association with poor treatment response and reduced disease-free survival. Transcription factor analysis traced elevated MMSDH expression to SP1, a transcription factor that binds the ALDH6A1 promoter in gemcitabine-resistant tumors, suggesting that the metabolic wiring of these cancers is rewired at the gene expression level before any drug ever enters the cell.

The mechanistic story then deepens in the oxygen-starved interiors of pancreatic tumors. Hypoxia, the researchers found, triggers an enzyme called GCN5 to install a lactyl group, a derivative of lactate, onto MMSDH at the amino acid lysine 113. This modification, known as lactylation, is part of a growing family of metabolite-driven protein modifications that have recently been shown to regulate DNA repair, chromatin biology, and now cancer metabolism. Lactylated MMSDH physically interacts with ACSL4 and, through its catalytic activity, generates propionyl-CoA, a short-chain acyl intermediate of valine catabolism. That propionyl-CoA is then handed to another acetyltransferase, KAT8, which uses it to attach a propionyl group to ACSL4 at lysine 606.

The consequences of this single chemical mark are profound. Propionylation at K606 repositions ACSL4 so that it binds HSC70, the cytosolic chaperone that recognizes the KFERQ-like targeting motifs required for chaperone-mediated autophagy, a selective degradation pathway in which individual proteins are unfolded and threaded into lysosomes. Once flagged in this way, ACSL4 is destroyed, its lipid-remodeling activity collapses, and the tumor cell becomes resistant to the lipid peroxidation that gemcitabine otherwise promotes. Using mass spectrometry, the team confirmed both the K113 lactylation on MMSDH and the K606 propionylation on ACSL4 in hypoxic pancreatic cancer cells, and showed that mutant versions of these proteins that cannot be modified fail to drive ACSL4 degradation or protect cells from ferroptotic death induced by gemcitabine, erastin, or RSL3.

Critically, the findings are not confined to cell culture. In an analysis of patient cohorts, tumors harboring high levels of MMSDH K113 lactylation and ACSL4 K606 propionylation displayed low ACSL4 protein, heightened ferroptosis resistance, and poor clinical response to neoadjuvant chemotherapy. Mouse xenograft experiments extended the picture: tumor cells engineered to express non-modifiable mutants of MMSDH or ACSL4 lost their protective shield and became acutely vulnerable to gemcitabine, while cells bearing the modification-enhancing wild-type enzymes grew aggressively even under low-oxygen conditions and responded poorly to treatment. The team also showed that the same axis operates in lung and ovarian cancer cell lines, hinting that MMSDH-mediated ACSL4 propionylation may be a broader mechanism of ferroptosis evasion across solid tumors.

Perhaps the most translational aspect of the study lies in two therapeutic strategies the authors developed to disrupt this axis. The first is dietary: because MMSDH is a valine catabolism enzyme that depends on valine-derived substrate to generate propionyl-CoA, restricting dietary valine in mice starved the pathway of its fuel. Combining valine-restricted diets with gemcitabine synergistically restored lipid peroxidation and suppressed tumor growth, notably without causing significant toxicity, weight loss, or metabolic distress in the animals. The second approach is pharmacological: the team designed cell-penetrating blocking peptides centered on the lactylated K113 sequence of MMSDH. A lead peptide disrupted the interaction between lactylated MMSDH and ACSL4, prevented ACSL4 propionylation and degradation, restored ferroptotic sensitivity, and markedly potentiated gemcitabine in both subcutaneous and orthotopic pancreatic tumor models while improving animal survival.

Beyond the immediate therapeutic implications, the study expands the conceptual map of how post-translational modifications couple cellular metabolism to cell fate. Amino acid catabolic enzymes are increasingly appreciated as moonlighting regulators of signaling and protein stability, and this work adds a new chapter by showing that a valine-processing enzyme can be co-opted by hypoxia-driven lactylation to flag a ferroptosis gatekeeper for lysosomal destruction. It also illustrates a chemical relay of remarkable economy: hypoxia produces lactate, lactate lactylates MMSDH, lactylated MMSDH produces propionyl-CoA from valine, and propionyl-CoA propionylates ACSL4, linking three metabolic programs, glycolysis, branched-chain amino acid catabolism, and lipid metabolism, into a single survival circuit. The authors’ clinical data suggest that MMSDH K113 lactylation and ACSL4 K606 propionylation may serve as biomarkers for predicting which pancreatic cancer patients will benefit from gemcitabine-based neoadjuvant regimens, potentially guiding treatment selection in a disease where therapeutic options remain painfully limited.

Challenges remain before this biology reaches the clinic. Blocking peptides must be optimized for delivery, stability, and specificity in humans, and dietary valine restriction will need careful evaluation in clinical trials, particularly given the catabolic state of many pancreatic cancer patients. Yet the study provides something pancreatic oncology has long needed: a mechanistically resolved, chemically validated, and clinically correlated explanation for why so many tumors shrug off chemotherapy, together with concrete tools to break that resistance. By exposing the GCN5-MMSDH-ACSL4 axis, the Zhejiang University team has turned a metabolic quirk of hypoxic tumor cells into a target, and in doing so opened a plausible path toward making ferroptosis-inducing chemotherapy a reality for one of medicine’s most intractable cancers.

Subject of Research: Hypoxia-driven MMSDH lactylation and ACSL4 propionylation as a mechanism of ferroptosis evasion and chemotherapy resistance in pancreatic ductal adenocarcinoma

Article Title: MMSDH facilitates ACSL4 propionylation to counteract ferroptosis upon hypoxia and impairs PDAC chemotherapy efficacy

Article References: MMSDH facilitates ACSL4 propionylation to counteract ferroptosis upon hypoxia and impairs PDAC chemotherapy efficacy. (n.d.). https://doi.org/10.1038/s43018-026-01236-w

Image Credits: AI Generated

DOI: 10.1038/s43018-026-01236-w

Keywords: pancreatic ductal adenocarcinoma, ferroptosis, MMSDH, ACSL4, lactylation, propionylation, hypoxia, gemcitabine resistance, chaperone-mediated autophagy, valine metabolism, GCN5, KAT8

The logic of the field has shifted because metabolites themselves turn out to be messengers. S-adenosylmethionine, or SAM, a product of one-carbon metabolism, serves as the methyl donor for chemical marks placed on mitochondrial DNA, including 5-methylcytosine and the bacteria-like N6-methyladenine. The methyltransferase METTL4 installs 6-methyladenine on mitochondrial DNA, which disrupts binding of the transcription factor TFAM and suppresses mitochondrial transcription. Hypoxia strongly induces METTL4 in liver and breast cancer cells, throttling mitochondrial metabolism precisely when tumors want it restrained. Methylation of the mitochondrial genome appears to protect it from oxidative damage, and DNA methyltransferases DNMT3A and DNMT3B mediate this in embryonic cells — enzymes whose aberrant expression or mutation drives cancer initiation. The picture is complicated, however: the true abundance of 5-methylcytosine on mitochondrial DNA remains debated, with optimized protocols reporting levels of only 0.19 to 0.67 percent in mouse oocytes and human lung carcinoma cells, and technical artifacts such as inefficient bisulfite conversion and mis-mapping of nuclear-integrated mitochondrial sequences continuing to muddy clinical analyses.

The replication machinery itself is equally metabolically entangled. DNA polymerase gamma, POLG, together with its accessory subunit POLG2, copies the mitochondrial genome, and its overactivation has been linked to elevated risk of prostate cancer and acute myeloid leukemia. Proteomic surveys have revealed a dense landscape of post-translational modifications on POLG and POLG2 that couple cellular fuel state to genome maintenance. The acetyltransferase GCN5 acetylates POLG at lysine 1039 using acetyl-CoA as donor, while the NAD+-dependent deacetylase SIRT3 reverses the mark — a push-pull system that reads acetyl-CoA and NAD+ availability directly. In parallel, poly(ADP-ribose) polymerase 1 uses NAD+ to attach ADP-ribose groups to POLG, triggering ubiquitination and degradation. Through these routes, the cell’s energy currency and redox state literally decide how much replication enzyme survives, and cancer cells exploit these metabolite-sensing nodes to rewire mitochondrial function to their advantage.

Transcription is governed by the same logic. The mitochondrial RNA polymerase POLRMT is succinylated at lysine 622, a modification that impairs its DNA-binding activity. The metabolic enzyme SUCLG1, part of the succinyl-CoA ligase complex, restricts mitochondrial succinyl-CoA levels to keep POLRMT hypo-succinylated, sustaining mitochondrial respiration and driving leukemia progression. Succinyl-CoA thus functions as a key metabolic signal bridging mitochondrial metabolism and transcription in both leukemia and solid tumors. TFAM, the packaging factor that recruits POLRMT to promoters, is itself phosphorylated, acetylated, and — most recently — arginine-methylated at position 82 by PRMT5, a modification essential for DNA binding and mitochondrial genome integrity. Loss of PRMT5 leaves breast carcinoma cells vulnerable to mitochondrial DNA damage and oxidative stress, and several PRMT5 inhibitors are already in clinical trials, raising the prospect of attacking tumor bioenergetics through the transcription apparatus itself.

Downstream, RNA processing and ribosome assembly consume enormous metabolic resources. The 12S and 16S ribosomal RNAs of the mitoribosome are studded with SAM-dependent methylation marks: TFB1M dimethylates the 3′ end of 12S rRNA to stabilize the small subunit, NSUN4 methylates cytosine 911, METTL15 installs N4-methylcytidine at position 839, and TRMT61B places N1-methyladenosine at position 947 of 16S rRNA, a modification conserved across all vertebrates that stabilizes local ribosome structure. Strikingly, when SAM is depleted, unprocessed pre-rRNA accumulates and mitoribosome assembly grinds to a halt. The same SAM pool feeds transfer RNA modifications, including the taurine-and-GTP-dependent wobble modification 5-taurinomethyluridine installed by GTPBP3 and MTO1, which is essential for translation fidelity — and, notably, MTO1 upregulation in melanoma sustains oxidative phosphorylation and confers resistance to vemurafenib. The dioxygenase ALKBH7, which depends on alpha-ketoglutarate and oxygen to erase methylation marks that otherwise delay transfer RNA maturation, adds another metabolite-gated checkpoint to the pipeline.

At the level of protein synthesis, the review highlights an increasingly granular view of mitoribosome regulation. Human mitoribosomes contain three iron-sulfur clusters installed by GLRX5 and BOLA3, enzymes with prognostic value in bladder cancer and lung adenocarcinoma, while METTL17 carries a [4Fe-4S] cluster that confers redox-sensing activity to protein synthesis. Lipid metabolism matters too: cardiolipin, enriched in cristae membranes where mitochondrial protein synthesis is spatially confined, promotes the inner-membrane docking of ribosomes, and the remodeling enzyme tafazzin maintains the membrane curvature this requires. Perhaps the most striking recent discovery is lactylation — the attachment of lactate-derived lactyl groups to proteins — now documented on TFAM and multiple ribosomal proteins in clinical tumor tissues, and linked to chemoresistance and metastasis in breast, kidney and lung cancers. Lactate, once dismissed as waste, is actively consumed by mitochondria as fuel and now appears to write epigenetic marks on the very machinery that builds the respiratory chain.

The mitochondrial genome also hides a second layer of coding information. Beyond its thirteen canonical proteins, it produces microproteins with potent biological activity. Humanin, a 24-amino-acid peptide, suppresses apoptosis by interfering with Bax activation; it is elevated in the serum of breast cancer patients and accelerates tumor growth, lung metastasis, glioblastoma invasion and temozolomide resistance in experimental models. MOTS-c, a 16-amino-acid peptide from the 12S rRNA open reading frame, activates AMPK and shows context-dependent effects — circulating levels rise in prostatic precancerous lesions and hepatocellular carcinoma yet fall in adrenal tumors, while in liver cancer MOTS-c promotes apoptosis under hypoxia and suppresses tumor growth. Additional peptides such as the SHLP family, SHMOOSE and CYTB-187AA underscore that the mitochondrial central dogma generates unanticipated regulators whose roles in cancer are only beginning to be mapped.

The translational payoff is substantial. The allosteric POLRMT inhibitor IMT1B blocks mitochondrial transcription with minimal toxicity in normal tissues and suppresses leukemia, endometrial carcinoma and other tumors in preclinical models. The thymidine analog alovudine inhibits POLG, impairing oxidative phosphorylation in acute myeloid leukemia, while small-molecule activators such as PZL-A allosterically restore mutant polymerase activity. FDA-approved antibiotics quinupristin and dalfopristin inhibit mitochondrial translation and reduce glioblastoma stem cell clonogenicity, and linezolid suppresses T cell metabolism through the same route, with hematological toxicities limiting its use. Inhibition of dihydroorotate dehydrogenase by ML390 starves leukemia cells of pyrimidines needed for mitochondrial biogenesis, and the iron chelator deferoxamine disrupts OXPHOS super-complexes in breast cancer with encouraging selectivity. Genome-wide sequencing has also revealed that colorectal cancer mitochondrial DNA mutations carry prognostic weight, with MT-CYB mutations significantly correlating with poor patient survival.

Perhaps the most futuristic front is mitochondrial genome editing itself. DddA-derived cytosine base editors and TALE-linked deaminases now allow precise CG-to-TA and A-to-G conversions in mitochondrial DNA, and adeno-associated virus vectors have delivered these editors to post-mitotic mouse tissue, while engineered mitoARCUS nucleases have cleared mutant mitochondrial DNA systemically in animal models. Off-target nuclear editing — reported at rates up to 17.5 percent at some sites — and unintended heteroplasmy shifts remain serious concerns, spurring strategies such as nuclear export signals, strand-selective editors like CyDENT, and tumor-specific promoters to restrict editing to cancer cells. The review’s authors close with a call to integrate high-resolution structural biology, single-cell metabolomics and in vivo lineage tracing to decode the spatiotemporal choreography of mitochondrial gene expression. If they are right, the organelle’s second genome — long overshadowed by the nucleus — may prove to be one of the most exploitable weaknesses a tumor carries.

evolutionary backdrop helps explain why this control system is so intricate. Mitochondria descend from a free-living prokaryotic ancestor engulfed more than 1.45 billion years ago, and the long history of metabolic trade-offs with the host cell left the organelle semi-autonomous: most of its roughly 1,500 proteins are nuclear-encoded, yet the thirteen genes retained in the compact circular genome encode core respiratory chain subunits whose expression must be tuned to local energy demand. Because mtDNA copy number varies naturally across tissues, its dysregulation is directly tied to cancer risk, and tumor cells dynamically adjust both genome abundance and gene output to match proliferative and metastatic needs.

Structural organization adds another layer of control. The mitochondrial genome packs into nucleoids within the matrix, with most protein- and rRNA-coding regions flanked by transfer RNA genes, and regulatory marks such as 5-methylcytosine concentrate in the displacement loop, the non-coding region where replication and transcription originate. Recent structural work has also complicated the classic picture of the replicase: POLG2 can contact DNA independently of POLG without altering polymerase activity, hinting at versatile functions beyond processive elongation.

Methodological caveats deserve emphasis for readers weighing the evidence. Detecting low-frequency mitochondrial modifications demands careful technique; antibody-based capture, bisulfite conversion, isotope labelling, HPLC–mass spectrometry and next-generation sequencing each carry distinct biases, and incomplete light-strand coverage can distort tumor measurements. As optimised protocols spread and quantitative data accumulate across tumor types, the field should be better placed to separate genuine metabolic regulation from technical noise, sharpening the therapeutic opportunities this review maps out.

Subject of Research: Metabolic regulation of the mitochondrial central dogma in cancer

Article Title: Exploring mitochondrial central dogma in cancer from a metabolic perspective

Article References: Zhang, S., Han, B., Chen, L., Chen, L., Wang, T., Lin, M., Liu, J., & Wang, Y. (2026). Exploring mitochondrial central dogma in cancer from a metabolic perspective. Clinical Cancer Bulletin, 5(1), Article 19. https://doi.org/10.1007/s44272-026-00071-5

Image Credits: AI Generated

DOI: 10.1007/s44272-026-00071-5

Keywords: mitochondria, mitochondrial DNA, mtDNA replication, POLRMT, TFAM, mitoribosome, RNA modifications, lactylation, mitochondrial translation, humanin, MOTS-c, mitochondrial genome editing