Is Depression Damaging Your Brain, or Just Pausing It?
Depression appears to press pause on the hippocampus's ability to make new neurons, not destroy it. A new Nature Medicine paper on brain tissue from unmedicated donors found the young progenitor cells still there, just held mid-lineage. The treatments already in the clinic, from SSRIs to esketamine to a regular running habit, look like the ones aimed at pressing play.
Brain imaging has long shown that depression comes with a slightly smaller hippocampus, the region that handles memory and mood. What that shrinkage meant was the argument. Killing neurons? Choking the machinery that makes new ones? Something else? A Nature Medicine paper from Peng and colleagues at Columbia, published August 21, 2026, gives the clearest answer so far. Working from hippocampal cells taken from unmedicated donors who died with major depressive disorder, they found the pipeline in the dentate gyrus wasn't empty. It was stalled.
Paused, not destroyed
The Columbia group paired single-nucleus RNA sequencing with chromatin accessibility and protein measurements, which is unusually thorough for postmortem tissue. Young progenitor cells were present. They weren't finishing the job. The cells carried a stress-related transcriptional signature and elevated interferon signaling, both consistent with a system being held in check rather than one that has been dismantled. A destroyed system needs rebuilding. A paused one needs something to unpause it.
That fits older work that never fully supported the idea that depression was killing brain cells. Postmortem stereology from Boldrini and colleagues has repeatedly shown that people with untreated depression had fewer mature dentate granule neurons and smaller dentate volume than controls, without a clear drop in progenitors. The progenitors were there. Something was blocking them from maturing and wiring in.
What SSRIs appear to restart
The same literature has a parallel finding. People with MDD who were on antidepressants at death had more neural progenitor cells, more small blood vessels feeding the neurogenic niche, and larger dentate volumes than untreated patients. In adult monkeys, blocking hippocampal neurogenesis with focal irradiation abolished the behavioral antidepressant effect of fluoxetine. A 2025 Translational Psychiatry study of 107 patients found that responders to escitalopram had larger left hippocampal volume at baseline and showed small volume increases in the right hippocampus over treatment that tracked with symptom improvement. Modest, and part-predictor and part-response, not a clean drug-induced growth spurt. Taken together, the treatments most people start with appear to act on the exact pipeline the new study finds paused.
That reframes the weeks-long lag before SSRIs kick in. Primate work puts adult dentate granule cell maturation on the order of at least five weeks, with some studies describing a program that runs for months. If the drug is nudging progenitors through a maturation window measured in weeks to months, the delay isn't the drug failing. It's the drug waiting for the neurons to grow up.
Faster levers already on the shelf
Not everyone can wait weeks, and not every antidepressant works for any given person. Newer options act on the same neuroplasticity machinery on faster timescales. Ketamine and its nasal-spray form esketamine drive rapid BDNF-mediated synaptic plasticity, and in January 2025 the FDA approved esketamine as the first monotherapy for treatment-resistant depression, a real category shift from its 2019 add-on approval. Lumateperone was cleared in November 2025 as an adjunct for MDD when standard antidepressants aren't enough. And moderate aerobic exercise produces a moderate-to-large antidepressant effect in randomized trials, competitive with first-line drugs, with the mechanism better established in animals than in humans.
| Option | Speed | Effect on hippocampal neuroplasticity | Best-supported use |
|---|---|---|---|
| SSRIs (e.g., fluoxetine, escitalopram) | Weeks to months | More progenitors, angiogenesis, and dentate volume in postmortem MDD tissue; modest hippocampal volume changes in escitalopram responders | First-line for moderate to severe MDD |
| Ketamine / esketamine (Spravato) | Hours to days | Rapid BDNF-mediated synaptic plasticity | Treatment-resistant depression; FDA monotherapy since Jan 2025 |
| Lumateperone (Caplyta) add-on | Weeks | Not directly established in humans | Adjunct to an antidepressant when response is partial (approved Nov 2025) |
| Moderate aerobic exercise | Weeks | BDNF upregulation supported in humans; dentate volume gains robust in animals, inconsistent in humans with depression | Adjunct across severity; low-risk, patient-preferred |
None of these work for everyone, and they aren't equivalent. But they attack the same problem from different angles, which is unusual for psychiatric treatments and unusually consistent with the biology the new study describes.
What the pause hypothesis can't prove yet
A single postmortem study, even a large and well-executed one, doesn't prove causation. The stalled signature could be a driver of depression, a consequence of it, or a marker of both. Older work on cell proliferation markers has been mixed. Exercise makes the translational gap concrete: in rodents it reliably grows dentate gyrus volume, while a dedicated trial in depressed patients found no hippocampal volume change despite improved fitness, and a 2025 meta-analysis found a null human effect in depressive contexts. There is no imaging test or blood marker yet that lets a clinician see this stall in a living person, which is what would make it directly actionable. Peng and colleagues narrow the interpretation. They don't settle it.
What would move the picture from strong mechanism to clinical tool is an in vivo marker of adult hippocampal neurogenesis paired with a trial showing that reversing the stalled signature tracks with remission. Until then, depression looks less like brain damage and more like brain silence, and the treatments already on the shelf are the ones that appear to break it.