Adrenomedullin rescues hypoxic interneuron migration

What the work claims

The authors use human forebrain assembloids and ex vivo developing human brain tissue to test whether hypoxia, a major risk factor for brain injury in extremely preterm infants, disrupts the migration of cortical interneurons. They report that mild hypoxia causes a rapid, reversible-like arrest of interneuron saltations without killing the cells, and that the hypoxia-induced peptide adrenomedullin is both necessary and sufficient to rescue migration. The rescue is mediated by RAMP2, the principal adrenomedullin receptor on interneurons, and depends on downstream cAMP/PKA signaling and GABA-A receptor expression. The central claim is that adrenomedullin is a hypoxia-responsive repair signal for interneuron migration and a candidate therapeutic target for preterm brain injury.¹

How it works

Cortical interneurons are born in the subpallium and migrate long distances into the dorsal forebrain during the second half of human gestation, precisely when preterm infants are exposed to hypoxic episodes. The authors model this with human forebrain assembloids made by fusing dorsal forebrain organoids, containing excitatory neurons, with subpallial organoids containing cortical interneurons tagged by a Dlxi1/2b::eGFP lentiviral reporter. Four hiPSC lines were used, two XX and two XY.

Live imaging on a confocal microscope with an environmental chamber switched cultures from 21% oxygen to less than 1% oxygen after 24 hours. The partial pressure of oxygen in the medium fell from roughly 150 mmHg to 25 to 30 mmHg. Hypoxia activated the expected response, stabilizing HIF1alpha and inducing hypoxia-responsive genes PFKP, PDK1, and VEGFA, but it did not increase cell death by Annexin V or cleaved caspase 3 staining.

The migration phenotype is specific. Interneurons move by repeated nuclear saltations. Under hypoxia the number of saltations fell by about 58%, and 70 of 129 actively migrating cells became completely stationary. Saltation length and directionality were unchanged, indicating that hypoxia does not disrupt the migration machinery itself but stops its repeated activation.

RNA sequencing of subpallial organoids identified 985 differentially expressed genes at 24 hours of hypoxia. The ADM gene had the highest fold change, with log2 fold change of 4.06 and an adjusted p-value of 6.08 times 10 to the minus 19. ADM protein also rose in the medium. Single-cell sequencing showed that ADM increases in all cell types but is especially high in ventral progenitors and astrocytes within subpallial organoids, while the receptor RAMP2 is preferentially expressed by interneurons and glutamatergic neurons and increases further in hypoxia.

Adding 0.5 micromolar human adrenomedullin during hypoxia restored saltation numbers to control levels. The effect was ligand-specific: denatured adrenomedullin, whose disulfide bond was alkylated with iodoacetamide, did not rescue, and the RAMP2 antagonist adrenomedullin 22-52 blocked rescue. Mechanistically, adrenomedullin increased intracellular cAMP, activated PKA, raised pCREB, and restored expression of GABA-A receptor subunits that were suppressed by hypoxia. Ex vivo slices of human cerebral cortex at about 20 post-conceptional weeks reproduced the 55% saltation decrease and ADM upregulation, extending the finding beyond the stem-cell model.¹

Where a skeptic should push

The most load-bearing assumption is that saltation arrest in a dish is the relevant cellular lesion in preterm brain injury. The authors show that hypoxia stops migration without immediate cell death, which is consistent with the idea that interneurons fail to reach their targets rather than dying outright. But whether the same mechanism operates in the intact preterm brain, where blood flow, inflammation, and multiple cell types interact, is still open.

The dose of adrenomedullin is pharmacological, 0.5 micromolar, chosen from prior in vitro work. It rescues migration acutely, but adrenomedullin has vascular, renal, and systemic effects that would complicate any neonatal therapeutic use. The paper does not test toxicity, optimal dosing, or delivery across the blood-brain barrier. The partial protection during reoxygenation also suggests the rescue is not complete: cells still migrate 35% less than controls after hypoxia ends, even if adrenomedullin was present.

Several controls strengthen the paper. The four hiPSC lines, the ex vivo fetal tissue validation, the ligand inactivation, and the receptor blockade all point to a specific RAMP2-dependent mechanism. Nevertheless, the assembloid model lacks immune cells, microglia, and the full vascular environment, so the findings are best viewed as a mechanistic proof of concept rather than a clinical recommendation.

Implication for brain organoid models of preterm injury

For organoid models of human organs, this paper advances the field in two ways. First, it establishes a quantitative live-imaging platform for human cortical interneuron migration, an event that is otherwise difficult to observe in human tissue. Second, it identifies a druggable signaling axis, adrenomedullin-RAMP2-cAMP/PKA, that can be probed in the same platform. That combination turns the assembloid from a descriptive developmental model into a functional assay for hypoxic injury and rescue.

The opportunity for drug discovery is clear. Preterm brain injury lacks effective therapies, in part because rodent models have repeatedly failed to predict human outcomes. A human assembloid assay that reports interneuron migration in real time could be used to screen for compounds that prevent or reverse hypoxia-induced migration arrest. Adrenomedullin itself is a peptide with established biology, so the path from hit to tool compound is more concrete than for a novel target. The readout, nuclear saltation frequency, is also objective and automatable.

The threat is premature translation. Adrenomedullin is a potent vasoactive peptide, and systemic administration to preterm infants would carry real risk. The assembloid data justify further mechanistic work and perhaps peptide-engineering approaches that restrict activity to the brain, but they do not yet justify clinical use. There is also a risk that the field over-interprets rescue in a reductionist model: restoring saltations in vitro is not the same as restoring interneuron positioning and circuit function in a developing brain.

The less obvious implication is for organoid validation. The authors use two human systems, assembloids and ex vivo fetal slices, and find the same phenotype. That is a useful template for the field. As brain organoids are increasingly used to model neurological disease, pairing them with primary human tissue where possible will be essential for claims that matter therapeutically.

The bottom line

This is a well-controlled primary study that identifies adrenomedullin as a hypoxia-responsive regulator of human cortical interneuron migration. The combination of live imaging, four hiPSC lines, ligand and receptor manipulations, and ex vivo fetal tissue validation makes the mechanism credible. What would confirm its therapeutic relevance is evidence that adrenomedullin or a RAMP2-biased agonist rescues interneuron positioning and circuit function in a more complex model of preterm brain injury, and that it can be delivered safely. What would weaken it is evidence that the migration arrest is an artifact of the assembloid format or that adrenomedullin's systemic effects override any central benefit. For now, it is best read as a strong mechanistic advance and a candidate drug-discovery starting point.