A clump of cells no bigger than a peppercorn, containing more than a million human cortical cells, kept firing electrical signals in laboratory culture medium for over five years. On August 19, 2026, a team led by Paola Arlotta (Golub Family Professor of Stem Cell and Regenerative Biology) at Harvard University published a paper in Nature titled "Human brain organoids record the passage of time over multiple years." The study showed that this long-term culture was more than just a new survival record—a "developmental clock" had been ticking inside the organoids all along (DOI: 10.1038/s41586-026-10877-x).
Co-author Arlotta calls this phenomenon a "warping of developmental time." When cells taken from an old organoid were transferred into a young organoid, the transplanted cells suddenly began producing late-stage neurons—ones that would normally appear only months later—even as the surrounding young cells were still generating early-stage neurons. It was as if the old cells "remembered" how much time they had lived through and leapt forward accordingly.
The wall organoids kept running into
Brain organoids are three-dimensional models of brain tissue grown from human pluripotent stem cells (iPS cells or ES cells). Since the mid-2010s they have spread rapidly as tools for observing development, building disease models, and screening drugs. Directly studying the living human brain is extremely difficult, both ethically and technically, and organoids have been expected to fill that gap.
However, this tool has faced two structural limitations.
The first is short lifespan. Neurons are especially fragile under culture conditions, making long-term survival difficult. In 2021, Aaron Gordon of UCLA and Sergiu Pașca of Stanford University reported in Nature Neuroscience a record of 694 days (about 23 months), the longest at the time. That study showed organoids reaching postnatal-equivalent stages by 250–300 days in culture, but it could not track maturation beyond that point.
The second is batch-to-batch variability. Even organoids made with the same protocol often differed greatly in cellular composition, making reproducible experiments difficult. Arlotta's team tackled this problem in a 2019 Nature paper, analyzing 166,242 cells from 21 organoids via single-cell RNA sequencing and demonstrating that 95% of the organoids reproduced nearly identical cellular diversity (Velasco et al., Nature 570, 523–527, 2019). This reproducibility protocol became the foundation for the current long-term culture work.
In other words, the reproducibility problem had been solved, but the lifespan problem remained. The human brain continues maturing for roughly 20 years after birth. Organoids that plateau after a few months cannot model postnatal processes such as synaptic pruning, myelination, and neural circuit reorganization. How to fill this gap was the starting point of the present study.
The key to survival: spontaneous firing and amino acids
The research team first tackled the factors preventing long-term neuronal survival. Prior studies had shown that spontaneous electrical firing in neurons improves survival rates. So the team introduced a specialized liquid medium designed to promote spontaneous firing, and added amino acid supplements as an energy source.
The effect was clear. Organoids cultured in this medium showed increased neuron numbers and denser synapses (connections between neurons) within nine months. At the one-year mark, every organoid grown in the new medium showed active electrical bursting activity, while none of the organoids in the control medium showed any activity at all. This electrical activity then persisted for at least two more years.
According to an NIH (National Institutes of Health) press release, the research team believes this medium was the key to long-term neuronal survival. However, a detailed breakdown of which components contributed how much remains a task for future work.
An epigenetic clock corroborates the "age"
Extending survival alone is not enough. The team needed to confirm that the organoids were actually maturing on the same schedule as the human brain. To answer this question, they integrated single-cell RNA sequencing, genome-wide methylation profiling, structural analysis, and electrophysiological recording.
The analysis covered 34 organoids collected at eight time points from 6 months to 5 years, combined with 51 organoids from previous datasets—110 organoids in total, comprising roughly 425,000 cells.
Among the results, the most decisive evidence came from an "epigenetic clock" based on DNA methylation. DNA methylation is a chemical modification that controls gene on/off states, and methylation levels at specific sites change in regular patterns with aging. The method of using this pattern to estimate the biological age of tissue was established by Steve Horvath and colleagues in 2013.
The team applied three types of epigenetic clocks (Horvath's pan-tissue clock, a human fetal brain-specific clock, and a human cortex-specific clock) to the organoids. The results showed that both the Horvath clock and the cortex-specific clock produced estimated DNA methylation ages that correlated strongly with actual time in culture (correlation coefficient r = 0.88–0.90). The median absolute errors were 7.25 months and 20.04 months, respectively.
This means that the organoids' molecular-level "age" closely matched the actual number of days in culture. Despite being cultured outside the body, the pace of maturation was just as slow as it is in the living human brain. This "slowness" itself appears to be a hallmark of a developmental program unique to the human brain.
| Item | Previous record (Gordon et al., 2021) | This study (Faravelli et al., 2026) |
|---|---|---|
| Longest culture period | 694 days (about 23 months) | Over 5 years (some specimens now at 7 years) |
| Analysis resolution | Bulk RNA sequencing, methylation arrays | Single-cell RNA sequencing + epigenetics + structure + function |
| Developmental stage modeled | Early postnatal (equivalent to ~250–300 days in culture) | Multi-year-scale postnatal maturation |
| Experimental verification of "memory of time" | None | Demonstrated via chimeric organoid transplantation experiments |
| Number of cells analyzed | Not specified in the paper | Approximately 425,000 cells (110 organoids) |
Testing the "memory of time" through transplantation
The fact that the epigenetic clock matches the organoid's actual age in culture is circumstantial evidence that organoids "record" time. But recording and remembering are different matters. To confirm whether cells actually "remember" their own elapsed time and whether this is reflected in their behavior, an intervention experiment was needed.
The research team took neural progenitor cells from organoids of different ages and mixed them together to create a single "chimeric organoid." This technique applies the "Chimeroids" method the same team reported in Nature in 2024 (Antón-Bolaños et al., Nature 631, 142–149, 2024).
When chemical signals promoting neuron production were given to the chimeric organoid, progenitor cells derived from the young organoid produced the neurons normally generated at early stages. Meanwhile, progenitor cells derived from the old organoid skipped the early stage entirely and directly produced late-stage excitatory neurons—the kind normally generated only months into culture. The old progenitor cells achieved production of these late-stage neurons, which typically takes more than two months, in just two weeks.
Even though the surrounding environment was signaling "we are still at an early stage," the old cells advanced according to their own internal clock. This result indicates that the developmental program is not driven solely by external environmental signals but also depends on time-related information recorded inside the cells.
Co-author Irene Faravelli (a postdoctoral fellow at Harvard University at the time of the research, now an associate professor at the University of Milan) said in the NIH press release: "The brain doesn't develop in a vacuum. It's a remarkably complex organ that interacts with many other systems. It was far from obvious that our simplified model would match natural development in so many respects."
The clock's identity remains unknown
The existence of a "cell-intrinsic developmental clock," as demonstrated by this study, offers one framework for explaining the slow maturation pace unique to the human brain. Previous research has suggested that epigenetic barriers forcibly slow the maturation pace of human neurons (Ciceri et al., 2024), and that species differences in mitochondrial metabolic rate influence developmental tempo (Casimir et al., 2024). However, the specific molecular mechanism driving the "recording and recall of time" demonstrated in this study has not yet been identified.
Organoids also lack a vascular system. It is known that once they exceed 500 micrometers in diameter, oxygen and nutrient supply becomes insufficient, causing necrosis at the core. This study partially overcame that problem through medium optimization, but careful interpretation is required regarding how far the maturation observed—achieved without an immune system, sensory input, or connections to other brain regions—corresponds to maturation in the living body.
In a Harvard press release, Arlotta said, "The goal is not to see how far we can possibly go," adding, "We now know we can go pretty far. The next thing to figure out is how to get there faster." The research team currently maintains organoids now in their seventh year, but they are not planning further culture aimed at breaking records.
The remaining questions converge on three directions: what molecules drive the hands of this clock, whether the clock's speed can be artificially altered, and how this ability to "leap through time" might be applied to disease modeling and regenerative medicine. A clump of cells that has kept time in a culture dish for five years has, for the first time, opened an experimental window onto the postnatal years that had long remained a "black box" in human brain development. What lies beyond that window, no one can yet see.