Lab-grown human brain tissue survived five years in a dish and matured on a human clock

A clump of human cortical brain cells, grown in a dish, has survived in culture for more than five years. The Harvard team behind the record reports the cells did more than endure. They began to mature along a human developmental schedule, picking up molecular features usually seen only after birth.

The work, published this week in Nature by Paola Arlotta's lab, is the longest systematic study of human brain organoids to date and the first to track them into what looks like postnatal development. The organoids are not miniature brains. They are microscopic spheroids of human cortical neurons, each smaller than a sesame seed, lacking blood vessels, immune cells, or any of the structures that make a brain a brain. The cultures gave researchers, for the first time at this timescale, a clean window into how human cortical cells age on their own.

The team profiled 34 organoids with single-cell RNA sequencing at eight time points between six months and five years, building a combined dataset of 110 organoids and roughly 425,000 individual cells. Each organoid contained more than a million cerebral cortex cells. As the cultures aged, their DNA-methylation patterns, the chemical marks that act as a molecular clock, shifted in step with the same clock measured in living human cortex. The cells were not just getting older in the dish. They were tracking human development.

The medium mattered. Standard brain-organoid culture recipes, designed to keep cells alive in a dish, run on sugar-rich formulations originally optimized for cancer cell lines. Arlotta's group modified a recipe called BrainPhys, lowering the glucose, adjusting ion concentrations toward physiological levels, and adding a glutamine-stability supplement. Neurons in older cultures stopped quietly dying and started firing in coordinated bursts. Network activity, the electrical chatter neurons use to talk to each other, held up past year two.

The lab's most striking experiment mixed cells from organoids of different ages. When nine-month-old neural progenitors were combined with fifteen-day-old cells, the older cells skipped ahead in developmental time, generating neurons that resembled two-month-postnatal human cortex in about two weeks. Arlotta called the effect a "time warp" in The Transmitter. The result points to an internal developmental timer that, once unlocked by a more permissive medium, runs on its own schedule.

The five-year number is itself a record, roughly three times the previous longevity mark of 694 days set by a UCLA/Stanford team in 2021, but Arlotta is clear that the field should not aim for longer cultures. "We're not going to be culturing organoids for 10 years," she told the Harvard FAS release. The paper positions the longevity work as the baseline that made shorter, faster protocols possible.

Independent experts see the same pivot. In-Hyun Park, who runs a brain-organoid program at Yale, framed the result as a way to model human postnatal development, not just fetal stages. Alysson Muotri, who directs a similar program at UC San Diego, called the five-year wait impractical and pushed for protein-level aging markers and weeks-long protocols. Muotri is co-founder of and holds equity in Tismoo; the conflict does not undercut the critique, but it shapes his standing in the field.

No third-party lab has yet replicated the five-year longevity claim, and the field's own definition of "matured" remains contested. Muotri's group, for example, uses electrophysiology and protein markers to track aging in organoids over weeks, not years, and reports features of postnatal neurons on a much shorter timeline.

The Harvard dataset is now a reference for future work. The single-cell atlas spans enough of human cortical development to anchor experiments: if a faster protocol can reproduce the same methylation and transcriptional signatures in weeks, researchers will have a way to compare the two side by side. The five-year cultures are the yardstick against which shorter protocols can now be measured.

The first authors, Irene Faravelli and Noelia Antón-Bolaños, have since moved to the University of Milan and University Medical Center Utrecht to start their own groups, a quiet footnote about how a multi-year longitudinal experiment gets handed off when it finally ends.