The research team began by asking a fundamental question about the egg itself. Oocytes, the precursors of eggs, carry their own DNA methylome, a landscape of methyl groups attached to cytosine bases that helps regulate which genes are active. When the researchers compared oocytes from mice fed a high-fat diet with oocytes from normally fed mice, they found nearly 940 differentially methylated regions, many of them sitting at genes with well-established roles in metabolism. Crucially, the changes were not random noise scattered across the genome. They clustered at specific regulatory loci, including regions controlling Hnf1α, a transcription factor that governs liver and pancreatic gene programs; Thra, encoding a thyroid hormone receptor involved in hepatic insulin sensitivity; and Pdk4, a kinase that suppresses glucose oxidation in favor of fat burning. These were precisely the kinds of genes whose misregulation could plausibly produce the metabolic phenotypes seen in offspring of obese mothers.

The next question was whether these methylation changes actually persist into the next generation, and here the study produced some of its most surprising findings. Embryos are known to undergo dramatic epigenetic reprogramming: shortly after fertilization, most DNA methylation across the genome is erased and then rebuilt anew, a process thought to wipe the slate clean between generations. The team tracked the methylation status of the obesity-associated regions through blastocysts, primordial germ cells and the tissues of first- and second-generation offspring. The methylation marks themselves were indeed erased during embryogenesis, exactly as the textbooks predict. Yet the genes did not return to normal. Hnf1α, Thra and Pdk4 showed persistent transcriptional dysregulation across two generations, and the methylation patterns re-emerged at the same loci in the germline of the next generation of females, suggesting that something other than the methyl marks themselves was carrying the memory forward.

That something, the researchers discovered, appears to be a histone modification. At the affected loci, enrichment of H3K36me2, a chemical tag on histone proteins that is known to recruit the de novo DNA methyltransferases in oocytes, was retained even after the DNA methylation had been stripped away. In other words, the embryo erased the methyl letters but kept the highlighted page. When the oocytes of the next generation were formed, the H3K36me2 signal guided the methylation machinery back to the same spots, re-establishing the aberrant pattern. This chromatin-based bookmark offers a mechanistic explanation for how environmental information can tunnel through the reprogramming barrier, and it aligns with earlier work showing that H3K36me2 and its trimerized cousin H3K36me3 form a platform essential for DNA methylation establishment in mouse oocytes.

To connect specific methylation changes to specific metabolic outcomes, the team turned to an elegant set of experiments using methylation editing. Rather than relying on diet to produce diffuse, genome-wide changes, they injected in vitro-matured oocytes with a catalytically dead Cas9 fused to the methyltransferase Dnmt3a, together with guide RNAs targeting single differentially methylated regions. This allowed them to hypermethylate one locus at a time and ask what each mark does on its own. The results were remarkably clean. Offspring derived from oocytes with targeted hypermethylation at the Hnf1α region showed enhanced hepatic gluconeogenesis, meaning their livers produced more glucose, while offspring with engineered hypermethylation at the Pdk4 region displayed impaired glucose tolerance. Both effects were female-biased, echoing the sex-specific patterns seen in the diet-induced model, and both persisted into the second generation when the edited females were bred onward.

The sex specificity is itself an important clue. Female offspring carried the metabolic consequences far more prominently than males, and methylation analyses across liver, kidney, brain and muscle showed that the aberrant marks were established in a tissue-specific pattern rather than uniformly across the body. The researchers also found that single CpG sites within the differentially methylated regions acted as regulatory landmarks correlating with gene expression, providing fine-grained markers that could eventually be used to read the epigenetic state of a locus without sequencing entire regions. This level of resolution matters because it moves the field from correlational observations, where obese mothers tend to have affected offspring, toward a causal, mechanistic account in which defined chemical marks at defined genes produce defined phenotypes.

Perhaps the most consequential finding for human medicine came when the team examined oocytes donated by women undergoing fertility treatment. In oocytes from women with obesity, the researchers detected conserved hypermethylation at the same HNF1A and THRA loci that had been flagged in the mouse studies. The convergence is striking: two species separated by tens of millions of years of evolution, exposed to obesity through very different diets and environments, show methylation changes at homologous metabolic genes. While the human sample cannot prove causation, the conservation of the pattern strongly suggests that the mechanism identified in mice is not a laboratory artifact but a feature of mammalian biology that operates in people.

The study builds on a growing body of evidence that the germline is a conduit for environmental memory. Earlier work has shown that sperm carry small RNAs that transmit acquired metabolic disorders to offspring, that paternal diet shapes offspring chromatin states, and that oocyte deficiencies in factors such as Stella and TET3 can transmit glucose intolerance from obese mothers to their young. What distinguishes the new research is its combination of genome-wide mapping across generations, identification of a chromatin-based persistence mechanism, and direct causal demonstration through targeted methylation editing. Together these threads turn a suggestive correlation into a coherent causal chain: maternal obesity alters the oocyte methylome at metabolic loci, H3K36me2 bookmarks those loci through embryonic reprogramming, methylation is re-established in the next generation’s germline, and the resulting gene dysregulation produces measurable metabolic dysfunction in offspring, preferentially in females.

The implications reach well beyond the laboratory. If the preconception oocyte is the critical window during which these marks are laid down, then interventions aimed at improving metabolic health before pregnancy, rather than during it, may carry particular weight for breaking the intergenerational cycle of obesity and diabetes. The finding that methylation marks are erased but re-established also suggests potential targets: if the H3K36me2 bookmarking step could be understood in finer detail, it might eventually become possible to interrupt the re-establishment of pathological marks without touching the normal reprogramming program. The authors caution that much work remains, including determining how broadly the mechanism applies across loci and confirming the human findings at scale. But the central message is already clear and, for a condition affecting hundreds of millions of people worldwide, sobering: the metabolic consequences of obesity can be written into the egg itself, in the language of DNA methylation, and read out again in the bodies of children and grandchildren.

Subject of Research: Intergenerational epigenetic inheritance of metabolic dysfunction via oocyte DNA methylation in maternal obesity

Article Title: Maternal obesity imprints methylation marks in oocytes to drive intergenerational metabolic dysfunction

Article References: Han, L., Wu, Y., Li, J., Chen, Y., Li, L., Wang, T., Chen, M., Wang, S., Li, C., Ding, G., Li, J., Sun, H., Shi, Z., & Wang, Q. (2026). Maternal obesity imprints methylation marks in oocytes to drive intergenerational metabolic dysfunction. Nature Metabolism. https://doi.org/10.1038/s42255-026-01617-6

Image Credits: AI Generated

DOI: 10.1038/s42255-026-01617-6

Keywords: maternal obesity, oocytes, DNA methylation, epigenetic inheritance, H3K36me2, Hnf1α, Pdk4, Thra, glucose metabolism, methylation editing, Nature Metabolism, intergenerational metabolic dysfunction

The research, led by Qianru Zhao, Bing Feng, and Vicky Dong of the Pennington Biomedical Research Center at Louisiana State University, together with colleagues at Baylor College of Medicine, the University of Illinois Chicago, Tulane University, Nanyang Technological University, and South-Central Minzu University, set out to answer a question that has puzzled neuroscientists for years. Pregnancy is known to reshape appetite—rodents eat more as gestation advances, and human studies have documented heightened craving frequency, emotional eating, and excess gestational weight gain. Yet while hypothalamic circuits governing hunger have been mapped in detail, the mechanisms behind the selective, sometimes obsessive desire for palatable foods during pregnancy remained largely unexplored at the level of individual neurons and their ion channels.

The team began by characterizing the behavior itself. Using the BioDAQ automated feeding-monitoring system, they gave female mice intermittent access to highly palatable diets—a high-fat diet, a high-protein diet, and a high-sucrose diet—and measured not just how much the animals ate but how hard they worked to seek food when it was inaccessible. On day fourteen of gestation, a stage comparable to the second trimester in humans, pregnant mice showed a striking escalation in craving-like behavior. They consumed larger meals of the palatable diets more frequently, spent more time probing the food hopper when food was withheld, and lingered longer in the hopper zone, all while their intake of ordinary chow and their body-weight gain remained comparable to those of virgin controls. The pattern, the authors note, echoes many of the behavioral signatures described in human pregnancy studies, from increased craving frequency to the dissociation between craving and caloric need.

With the behavioral phenotype established, the researchers turned to the brain. The dorsal raphe nucleus, a narrow ridge of cells along the brainstem’s midline, supplies the bulk of the brain’s serotonin, a neurotransmitter long implicated in mood, appetite, and reward. Using whole-cell patch-clamp electrophysiology, the team recorded from fluorescently identified serotonin neurons in the dorsal raphe of virgin mice and pregnant mice at gestational day fourteen. The result was unambiguous: serotonin neurons from pregnant animals fired action potentials at significantly lower frequencies than those from virgin animals, and their resting membrane potentials reflected a hyperpolarized, less excitable state. Notably, the suppression persisted into the postpartum period, hinting that the change is not a fleeting response to a single hormonal moment but a sustained rewiring of the cell’s electrical properties.

To understand why the neurons fell silent, the investigators turned to single-cell transcriptomics. Patch-seq analysis, which combines electrophysiological recording with single-cell RNA sequencing, revealed a molecular shift in the serotonin neurons of pregnant mice: increased expression of the gene encoding the small-conductance calcium-activated potassium channel type 3, or SK3. These channels open in response to calcium influx during action potentials, allowing potassium ions to flow out of the cell and clamping the membrane back toward its resting potential. More SK3 current means each spike is followed by a stronger repolarizing brake, effectively throttling the neuron’s firing rate. The team confirmed the upregulation at the protein level, showing stronger SK3 immunofluorescence in serotonin neurons of pregnant animals, and measured larger apamin-sensitive SK currents—the pharmacological fingerprint of SK3 activity—in recordings from pregnant mice.

The causal test followed. When the researchers genetically deleted SK3 channels selectively from dorsal raphe serotonin neurons, two things happened. First, the pregnancy-associated suppression of firing frequency vanished; the neurons of pregnant mice lacking SK3 fired as briskly as those of virgins. Second, and more strikingly, the food-craving-like behavior of the pregnant animals dropped. Mice without SK3 in their serotonin neurons no longer showed the exaggerated seeking, consumption, and hopper-directed persistence of their pregnant wild-type counterparts. The implication is direct: the quieting of serotonin neurons, mediated by SK3, is not a byproduct of pregnancy cravings but a driver of them.

The converse experiment sealed the argument. In virgin female mice—animals that would not normally crave—the team overexpressed SK3 in dorsal raphe serotonin neurons using viral vectors. The manipulation suppressed neuronal firing and, remarkably, reproduced the pregnancy phenotype in animals that had never been pregnant. These virgin mice began to eat more palatable food, seek it more often, and spend more time at the hopper, mimicking the behavior of gestational-day-fourteen dams. A single ion channel, in a single population of brainstem neurons, was sufficient to install a craving-like state in a non-pregnant brain.

But where do these serotonin neurons act to change behavior? The answer pointed to the mesolimbic reward system. Anatomical tracing and optogenetic experiments showed that dorsal raphe serotonin neurons send monosynaptic projections to the ventral tegmental area, the hub of the brain’s dopamine reward circuitry. When the team activated this DRN-to-VTA pathway with channelrhodopsin, food-craving-like behavior in female mice diminished; when they inhibited the projection with halorhodopsin, the behavior was modulated in the opposite direction. Chemogenetic activation of the same circuit in pregnant mice reduced their craving for palatable diets. Intriguingly, stimulating serotonin terminals in the nucleus accumbens or lateral hypothalamus—two other major serotonin projection targets—failed to alter craving behavior, sharpening the picture of a circuit specifically routed through the ventral tegmental area. The emerging model is elegant: during pregnancy, SK3 channels silence serotonin neurons, releasing their inhibitory grip on the VTA’s reward machinery and thereby amplifying the motivational pull of palatable food.

The findings arrive at a moment of growing clinical concern about gestational weight gain. Prior research has shown that food-craving frequency during pregnancy mediates the relationship between emotional eating and excess weight gain, and that accumbal dopaminergic circuits mediate craving-like episodes in pregnant mice—work from a 2022 Nature Metabolism study that the present results build upon and extend by identifying an upstream serotonergic control point. Excessive gestational weight gain is associated with maternal obesity, gestational diabetes, and long-term metabolic risk for both mother and child, yet safe and effective interventions remain scarce. By pinpointing SK3 channels in dorsal raphe serotonin neurons as a molecular switch, the study offers a concrete pharmacological target. SK channels are already the focus of drug-development efforts for other neurological conditions, and modulators of serotonergic signaling are among the most clinically mature tools in neuropsychiatry. Translating the mouse findings to humans will require caution—rodent gestation differs from human pregnancy in important hormonal and neuroanatomical respects, and optogenetics cannot be applied to patients—but the logic of the circuit suggests that restoring serotonergic tone, or dampening SK3 activity, might temper pathological cravings without suppressing the healthy appetite increase that pregnancy demands.

Beyond the clinical horizon, the study reframes a familiar human experience. The sudden, almost gravitational attraction to sweets and fats that so many pregnant women describe may not be a psychological weakness or a trivial side effect of hormonal flux, but the output of a precisely engineered neural adaptation—one that evolution may have favored to ensure developing offspring receive energy-dense nutrition. As Zhao, Feng, Dong, and their colleagues demonstrate, that adaptation can be traced to the opening probability of a potassium pore in a handful of brainstem neurons, and it can be switched on in a non-pregnant brain by forcing that pore open, or switched off in a pregnant one by deleting the gene that builds it. Few stories in modern neuroscience connect a molecule, a circuit, and a deeply human behavior so cleanly. The next chapter—determining whether the same serotonergic brake governs cravings in pregnant women, and whether it can be safely tuned—will be watched closely by neuroscientists, obstetricians, and anyone who has ever wondered why the pickle jar suddenly seems irresistible.

Subject of Research: Serotonin neuron control of pregnancy-induced food craving behavior via SK3 ion channels

Article Title: Serotonin neurons in the dorsal raphe control food-craving-like behavior during pregnancy in mice

Article References: Zhao, Q., Feng, B., Dong, V., Lau, L. H., Liu, H., Yu, M., Liang, K., Tran, C., Feng, H., Smiley, T., Gao, P., Yan, A., Ye, H., Jiang, Y., Wang, C., Xu, P., & He, Y. (2026). Serotonin neurons in the dorsal raphe control food-craving-like behavior during pregnancy in mice. Nature Neuroscience. https://doi.org/10.1038/s41593-026-02445-3

Image Credits: AI Generated

DOI: 10.1038/s41593-026-02445-3

Keywords: serotonin neurons, dorsal raphe nucleus, food cravings, pregnancy, SK3 ion channel, ventral tegmental area, mice study, Nature Neuroscience, maternal obesity, neuroscience, appetite regulation, reward circuitry