Cellular memory allows information about identity, lived experiences and received stimuli to be preserved. At the PRBB centres, different teams study how these molecular memories are formed and how they can protect us or contribute to disease.
We often think of memory as a faculty exclusive to the brain. But the past also leaves a trace at a microscopic scale. A skin cell preserves its identity when it divides. Some immune cells allow the body to respond more quickly to a pathogen it has already encountered. And a tumour cell may respond differently to a drug after having been exposed to it. In all three cases, we are talking about memory, although the mechanisms are not the same.
At the centres of the Barcelona Biomedical Research Park (PRBB), several teams are trying to understand how cells preserve information about what they are, what they have experienced and how they have responded. They do so through mathematical modelling, epigenetics, molecular epidemiology, immunology and single-cell technologies. The questions vary, but they all start from the idea that the behaviour of a cell depends not only on its DNA and on what surrounds it at that moment, but also on its history.
Key points
- In biology, we talk about cellular memory when a past experience leaves a persistent change that conditions a cell’s future response.
- Cells can preserve information through changes in chromatin, marks on DNA or histones, networks of proteins and RNA, or the persistence of specific cell populations.
- These mechanisms help maintain identity and support adaptation, but they can also favour ageing, treatment resistance or relapse.
What does it mean to remember without a brain?
Cells do not store images, nor do they have a specific place where memories are archived. In biology, the word “memory” is used to describe a different situation. Cellular memory refers to the fact that a past experience can modify the state of a cell, and that this change can persist even once the initial stimulus has disappeared. As a result, if the same situation occurs again, the cell may respond in a different way.
This persistence may last minutes, months or, in some cases, be passed on to daughter cells when the cell divides. It can also be preserved in different ways at the molecular scale. Information may be stored in the organisation of chromatin, in chemical marks on DNA or histones, or in protein and RNA circuits. In the case of the immune system, by contrast, memory depends on the persistence of specialised cell populations.
Therefore, there is no single mechanism of cellular memory. To be able to talk about it, it must be shown that the past experience has genuinely modified the way the cell responds in the present.
Rosa Martínez-Corral, who leads the Theoretical Regulatory Biology group at the Department of Medicine and Life Sciences at Pompeu Fabra University (MELIS-UPF), investigates these boundaries using mathematical models. In 2024, she co-led a study, published in Current Biology, which explored how an individual cell could show habituation, a simple form of learning in which the response decreases in the face of a repeated stimulus.
The team analysed which biochemical networks could reproduce the characteristics of habituation. The models combined fast and slow processes. The former allowed an immediate response to a signal, while the latter preserved information about previous exposures and adjusted the following response. This does not mean that a cell learns in the way a brain does, but it does show that a relatively simple molecular network could modify its behaviour according to what has happened before.
To talk about cellular memory, it must be shown that a past experience modifies the present response.
More recent work from the Institute of Evolutionary Biology (IBE: CSIC-UPF) has experimentally tested another form of cellular learning. In 2026, Maor Knafo, Elena Casacuberta and Iñaki Ruiz-Trillo published in PRX Life a study with Capsaspora owczarzaki, one of the closest unicellular relatives of animals.
The team exposed the cells to a light and vibration signal that predictably preceded moderate heat stress. A control group received the same stimuli, but in a random order. After 35 training cycles, the cells faced a heat shock of 38 °C. Mortality was 12,1% among cells trained in the predictable environment and 26,8% in the control group.
The authors interpret these results as evidence of associative learning and adaptive memory within the same generation. This does not mean that cells learn like a brain, but it does mean that they can link previous signals with a future situation and adjust their response. The molecular mechanism that preserves this information is not yet known.
Remembering who you are after dividing
The cells in our body contain practically the same DNA, but each cell type activates specific genetic programmes and keeps others silent. For example, a neuron activates the genetic programmes that allow it to transmit signals, while a skin cell expresses the genes needed to form a protective barrier.
In addition, when one of these cells divides, the daughter cells must maintain this genetic pattern in order to remain the same cell type. It is not enough to copy the DNA sequence: the cell must also reconstruct which regions of the genome should remain accessible and which should continue to be silenced.
This is where epigenetics comes into play, that is, the study of the mechanisms that regulate gene expression without altering the DNA sequence. DNA methylation, histone modifications, the three-dimensional organisation of the genome and different regulatory proteins contribute to maintaining a cell’s genetic pattern. Together, these epigenetic mechanisms allow cellular identity to be preserved without changing the letters of the DNA.
Sergio Aranda, a researcher in Luciano Di Croce’s laboratory at the Centre for Genomic Regulation (CRG), studies how cells regulate chromatin to preserve this identity and guide their fate. In 2025, he took part in a study published in eLife that analysed how heterochromatin, the most compact part of the genome, is organised when embryonic stem cells move from a state similar to that of the two-cell embryo to the pluripotent state. The team identified two proteins, SMARCAD1 and TOPBP1, that contribute to restoring and maintaining this organisation. Although the study does not directly analyse the transmission of memory between divisions, it helps us understand how a cell reconstructs the epigenetic state associated with its identity.
This memory must be stable, but not rigid. Stem cells must preserve what defines them while also maintaining the ability to give rise to other cell types. The balance between stability and plasticity is essential for development and tissue regeneration. When it fails, a cell may activate a genetic programme at the wrong time or in the wrong place.
The trace of what we have experienced
Identity is not the only trace a cell can preserve. Tobacco smoke, pollution, diet and other environmental exposures are associated with changes in DNA methylation, gene expression, proteins or metabolism. These marks can help us reconstruct part of what the organism has experienced.
Mariona Bustamante and her team, from the Barcelona Institute for Global Health (ISGlobal), study interactions between genes and the environment, as well as the molecular mechanisms that may connect exposures with health. In a study led by ISGlobal and published in Environment International in 2025, the team analysed blood samples from 2.695 children aged 7 to 10 from eight European countries. They found changes in DNA methylation in 11 regions associated with exposure to tobacco smoke at home.
These results show that an exposure can leave a detectable molecular imprint. However, this does not necessarily prove that each cell “remembers” smoke or that the observed changes cause disease. Epigenetic analyses make it possible to identify associations, but it is still necessary to determine whether the changes have a biological function, how long they persist and to what extent they directly reflect the exposure. For this reason, in molecular epidemiology it is often more precise to speak of an imprint or biomarker than of functional memory.
Not all molecular imprints are functional memory.
When the past modifies the future response
The most intuitive function of memory is to prepare us for what may happen again. The immune system offers the best-known example. After an infection or a vaccine, some B and T cells persist for a long time and retain receptors capable of recognising specific antigens. If the pathogen reappears, these populations respond more quickly and intensely. Immune memory therefore does not work in the same way as the epigenetic memory that maintains a cell’s identity. What they have in common, however, is that a past experience conditions the future response.
The same capacity for adaptation can also become a problem. In cancer, some cells survive a treatment and give rise to a population that responds differently when exposed to the same drug again. Sometimes, this persistence is related to epigenetic changes or reversible cellular states. In other cases, resistance appears because the treatment selects cells that already had mutations or characteristics that made them less vulnerable.
Anna Bigas, who leads the Stem Cells and Cancer group at the Hospital del Mar Research Institute, studies how genetic and epigenetic diversity contributes to the persistence of tumour cells. In 2026, the group launched a project on T-cell acute lymphoblastic leukaemia and the alterations affecting the BAF complex, which is involved in chromatin remodelling. The team will compare cellular states before and after exposure to chemotherapy to identify which characteristics allow some cells to survive and contribute to relapse.
Some cellular memories could be reversible. This does not mean that we can erase them easily or selectively. The same mechanisms that maintain a harmful state may also be essential for preserving the identity and normal functioning of the cell.
The same mechanisms that maintain a harmful state may also be essential for the normal functioning of the cell.
Cells do not keep a diary of what has happened to them. They preserve molecular states built from their history. These states allow them to maintain their identity, adapt to what they have experienced and prepare the next response. Sometimes they protect the organism. At other times, they contribute to ageing, treatment resistance or disease.
Understanding cellular memory also requires distinguishing which memories should be preserved, which ones we might be able to modify and which imprints we still do not know how to interpret. This distinction will help us consider how to intervene in a harmful memory without altering the information the cell needs to continue functioning.
Find out more
- Theoretical Regulatory Biology Group, led by Rosa Martínez-Corral (MELIS-UPF).
- Multicellgenome Lab, led by Elena Casacuberta and Iñaki Ruiz-Trillo (IBE, CSIC-UPF).
- Di Croce Lab – Epigenetic Events in Cancer, led by Luciano Di Croce (CRG).
- Omics and bioinformatics, coordinated by Juan Ramón González and Mariona Bustamante (ISGlobal).
- Stem cells and cancer, led by Anna Bigas (HMRIB).
Eckert, L., Vidal-Saez, M. S., Zhao, Z., Garcia-Ojalvo, J., Martinez-Corral, R. & Gunawardena, J. (2024). Biochemically plausible models of habituation for single-cell learning. Current Biology, 34(24), 5646–5658.e3. doi: 10.1016/j.cub.2024.10.041.
Knafo, M., Casacuberta, E. & Ruiz-Trillo, I. (2026). Beyond Diffusion: Bayesian Learning Strategies in Single-Cell Life. PRX Life, 4(2), 023025. doi: 10.1103/5p5z-t8qy.
Sebastian-Perez, R., Nakagawa, S., Tu, X., Aranda, S., Pesaresi, M., Gomez-Garcia, P. A., et al. (2025). SMARCAD1 and TOPBP1 contribute to heterochromatin maintenance at the transition from the 2C-like to the pluripotent state. eLife, 12, RP87742. doi: 10.7554/eLife.87742.3.
Cosin-Tomas, M., Hoang, T., Qi, C., Monasso, G. S., Langdon, R., Kebede Merid, S., et al. (2025). Association of exposure to second-hand smoke during childhood with blood DNA methylation. Environment International, 195, 109204. doi: 10.1016/j.envint.2024.109204.
