Electroacupuncture protects mouse brains from stroke via lactylation-driven ZIP14 expression

Needles, Lactate, and the Epigenome: Electroacupuncture Blocks an Iron Cascade That Kills Stroke-Vulnerable Neurons

For centuries, electroacupuncture has sat uneasily between traditional practice and hard neuroscience: clinically promising, mechanistically opaque. A new study now traces one of its effects all the way down to the chemical decoration of DNA-packaging proteins, exposing an unexpected chain of events by which stroke damage spreads through the brain. Writing in the Journal of Advanced Research, a team at the Fourth Military Medical University in Xi’an, China, reports that electrical stimulation of a single acupuncture point on the scalp lowers the surge of lactate that follows a stroke, blocks a lactate-derived histone mark from switching on an iron-importing gene, and ultimately shields neurons in the ischemic penumbra, the fragile borderland of tissue that survives the initial insult, from ferroptosis, an iron-fueled form of regulated cell death. The work, anchored by a randomized clinical trial and an unusually deep molecular dissection in mice, lands just as lactylation, the discovery that metabolites can chemically edit histones, migrates from cancer biology into neuroscience, and it hands drug developers a fresh target in one of medicine’s most stubborn problems.

The stakes are considerable. Ischemic stroke remains a leading global cause of death and permanent disability, and standard care, meaning intravenous thrombolysis and mechanical thrombectomy, is built to restore blood flow fast. Yet fewer than 30 percent of patients benefit from thrombolytic treatment, hemmed in by hemorrhage risk and a narrow therapeutic time window. Worse, reperfusion itself can inflame the damage: a phenomenon known as cerebral ischemia/reperfusion injury (CIRI) accelerates neuronal death in the penumbra even after circulation returns. Adjunctive therapies that could salvage these surviving neurons are urgently needed. Acupuncture and its electrified variant have shown efficacy against ischemic stroke in both clinical and animal studies, with reported effects ranging from reduced inflammation to inhibited pyroptosis and enhanced autophagic flux and synaptic plasticity, but the fundamental processes have remained poorly understood, a knowledge gap that has hindered optimization and wider clinical application.

To anchor the mechanism in humans, the team ran a randomized, blinded controlled trial in 28 patients scheduled for endovascular intervention for stroke under general anesthesia. Participants were assigned 1:1 to electroacupuncture at GV20 (Baihui), the point where the midline of the skull meets the line connecting the peaks of the ears, or to sham needling without electrical stimulation, applied from anesthesia induction to the end of surgery. Patients, anesthesiologists, and outcome assessors were blinded to allocation. The primary endpoint, arterial lactate measured 30 minutes after recanalization, fell significantly in the EA group: 1.0 ± 0.3 mM versus 1.4 ± 0.3 mM in controls, a mean difference of 0.4 mM (95% CI 0.2–0.5). Arterial pH, pCO2, glucose, base excess, and bicarbonate were unchanged, pointing to a specific metabolic effect rather than a general shift in acid–base balance. Follow-up modified Rankin scale scores were similar between the groups, a null at this sample size that leaves longer-term functional benefit an open question.

The mechanistic campaign in mice supplied what the trial could not. In a model of transient middle cerebral artery occlusion and reperfusion (MCAO/R), five consecutive days of pretreatment with electroacupuncture, 30 minutes daily at 1 mA with a dense-disperse 2/10 Hz waveform at GV20, shrank infarct volumes on TTC-stained brain sections and improved modified neurological severity scores at 24 hours. Behavioral gains were not trivial: treated animals clung to an accelerating rotarod measurably longer and gripped harder with their forelimbs than stroke controls one day after reperfusion. Critically, the therapy also worked when delivered after the stroke, at 6 and 18 hours post-reperfusion, mirroring how acupuncture is actually administered in the clinic. Post-stroke EA downregulated PKM2, cut lactate accumulation, and suppressed ferroptosis, and it still mitigated brain injury when the protection was challenged by pharmacological reactivation of PKM2.

At the pivot of the mechanism sits pyruvate kinase M2 (PKM2), the glycolytic enzyme increasingly described as a “metabolic switch” that determines neuronal fate through both metabolic and nuclear actions. The brain, the body’s greediest consumer of oxygen and glucose, pivots from oxidative phosphorylation to glycolysis when perfusion fails, and lactate levels rise in stroke patients and animal models alike. The researchers found that CIRI elevated both the abundance and the enzymatic activity of PKM2, increasing its levels in the cytoplasm and the nucleus of penumbral neurons, and that EA reversed both. RNAscope probing and immunostaining confirmed that Pkm2 mRNA and protein sit predominantly in neurons, and the mRNA itself did not change, hinting at post-transcriptional control of the enzyme’s activity and localization. To determine which arm of PKM2 mattered, the team used ML-265, a compound that activates PKM2’s metabolic function while blocking its nuclear translocation. In EA-treated mice, ML-265 raised lactate and histone lactylation and erased the neuroprotection, whereas a viral construct engineered to preferentially boost nuclear PKM2 left infarct volumes and behavior essentially untouched, evidence that at this early post-reperfusion stage, PKM2’s metabolic output, not its nuclear signaling, carries EA’s benefit.

The next question was where all that lactate goes. Beyond serving as an emergency fuel, lactate is a substrate for lactylation, a recently discovered post-translational modification in which a lactyl group is chemically coupled to lysine residues, most consequentially on the histone proteins that package DNA. Histone lactylation loosens chromatin and switches on transcription; it is well documented in cancer metabolism but poorly mapped in neurological disease. Pan-lysine lactylation rose globally after injury and was suppressed by EA, but a site-specific screen sharpened the signal: of six residues tested, only lactylation at histone H4 lysine 12 (H4K12la) rose reliably in the ischemic penumbra, while H3K9, H3K18, H4K5, H4K8, and H4K16 were unchanged. Immunofluorescence costaining localized the surge almost entirely to NeuN-positive neurons, with only slight changes in astrocytes and microglia, and EA reversed it.

To find the genes under H4K12la’s control, the researchers stacked three genome-wide assays: Cleavage Under Targets and Tagmentation (CUT&Tag) to map where the mark lands, ATAC-seq to chart chromatin accessibility, and RNA-seq to quantify transcription. Stroke opened roughly 64 percent of 29,070 differentially accessible regions and shifted expression of 2,896 genes, with the 171 genes that were both chromatin-opened and transcriptionally upregulated enriched for ferroptosis pathways; promoter analysis also flagged AP-1, the inflammation-linked transcription factor complex, hinting at crosstalk between metabolism and ischemic inflammatory gene regulation. CUT&Tag then identified 39 peaks of differential H4K12la, 92 percent of them increased after injury, and intersecting all three layers left just four genes that simultaneously gained the mark, chromatin opening, and elevated transcription. The standout was Zip14, which encodes ZIP14, a broad-scope metal-ion transporter that imports zinc, manganese, and non-transferrin-bound iron. ChIP-qPCR confirmed that H4K12la enrichment at the Zip14 locus far exceeded that of H3K18la in neurons, was minimal in astrocytes, and was absent in microglia, and immunostaining showed EA reversing the neuronal ZIP14 upregulation.

The consequences were ferroptotic. Overexpressing ZIP14 in HT22 neuronal cells raised intracellular Fe2+ and malondialdehyde, a lipid-peroxidation product. In mice, CIRI depleted the antioxidants glutathione and superoxide dismutase, elevated malondialdehyde, cut GPX4, the glutathione peroxidase that serves as ferroptosis’s master brake, and ravaged mitochondria, whose outer membranes ruptured and cristae vanished under the electron microscope. The GPX4 loss localized to neurons, while ACSL4 and LPCAT3, two lipid-remodeling enzymes often tied to ferroptosis, were unchanged, sharpening the picture of an iron-import-driven death program. EA reversed every one of these signatures. The team then stress-tested the axis from both ends. Injecting sodium lactate into EA-treated mice restored H4K12la, revived ZIP14 and TFR1 expression, worsened mitochondrial damage, and enlarged infarcts with poorer behavioral scores. Conversely, deferoxamine, an iron chelator already tested in human clinical trials, rescued mice whose EA protection had been dismantled by ML-265, restoring smaller infarcts and better motor performance. Together, the experiments established a causal PKM2–lactate–H4K12la–ZIP14–ferroptosis pathway as the operative target of electroacupuncture.

The findings braid together three of the hottest threads in modern biology, metabolic reprogramming, epigenetics, and regulated cell death, and they resolve, in part, a long-standing paradox. Lactate was once considered a potential therapy for stroke, with evidence that it can serve as an alternative energy substrate, enhance cerebral blood flow, and dampen neuroinflammation; more recent work showed it can aggravate neuronal death through lactylation of non-histone proteins. This study identifies histone lactylation as the missing link to ferroptosis and suggests that lactate’s pleiotropic effects in the brain depend on dose, timing, route, and cell type. It also reframes ferroptosis itself, pointing to central carbon metabolism rather than lipid metabolism alone as the ignition source, a view consistent with recent reports that histone lactylation drives ferroptotic death in sepsis-associated lung injury and myocardial ischemia–reperfusion. The authors flag a practical corollary too: poorly controlled diabetes raises stroke risk 1.5- to 3-fold, implying that restricting dietary sugar might modestly aid stroke control. And because ZIP14 mutations in humans cause rapidly progressive childhood-onset parkinsonism-dystonia, H4K12la now offers a transcriptional handle on ZIP14-related diseases well beyond stroke.

Limitations temper, but do not dismantle, the story. The mechanistic snapshot rests on a single 24-hour post-reperfusion time point, with markers behaving inconsistently at 6 and 12 hours and normalizing by three days, hinting that both injury and protection may be phase-dependent. Histone H4 is encoded by many genes, making H4K12 point-mutant mice infeasible, and the writers, erasers, and readers of the mark, with candidates including p300, Brg1, HBO1, KAT8, and HDAC1-3, remain to be identified in this setting. PKM2 is expressed beyond neurons, so astrocytic and microglial contributions cannot be excluded, and the trial’s perioperative timing of EA differs from the mouse pretreatment protocol. Even so, the study delivers something rare: a mechanistic account of acupuncture precise enough to be measured, mimicked, and improved, whether by tuning stimulation parameters, pairing the therapy with iron chelators or ferroptosis inhibitors, or using lactate and PKM2 as biomarkers to personalize treatment. An ancient practice, viewed through the lens of chromatin chemistry, turns out to be quietly editing the genome’s instructions for neuronal life and death.