Barcelona team develops AI that detects ageing in blood stem cells

Ageing gradually undermines the body's ability to regenerate, severely affecting the production of blood cells. In this race against the biological clock, analysing changes in blood stem cells is crucial to opening the door to interventions that can slow their decline and restore their original function. ADVERTISEMENT ADVERTISEMENT A scientific team in Barcelona has developed ChromAgeNet, an artificial intelligence tool that detects ageing in blood stem cells by analysing the three-dimensional structure of their DNA. Published in the journal "Aging Cell", the research, led by Maria Carolina Florian (IDIBELL) and Paula Petrone (BSC-CNS/ISGlobal), shows that the organisation of chromatin contains information that makes it possible to distinguish cells from young and aged mice. A bridge between biomedicine and supercomputing To train the model, the researchers applied convolutional neural networks to 3D images of the nuclei of mouse stem cells (source in Spanish). ChromAgeNet uses neural networks to decipher how DNA ages. AI makes it possible to detect subtle structural variations that are imperceptible to the human eye. One of the main concerns when using neural networks in biomedicine is not knowing what they base their decisions on. ChromAgeNet overcomes that barrier by offering an explainable AI (XAI) system. The tool managed, on average, to correctly classify around 68% of cells as coming from young or aged mice using images obtained with DAPI, a standard, low-cost stain. In addition, ChromAgeNet stands out as an explainable AI, allowing researchers to investigate which image features contribute to its predictions. These analyses highlighted, among other regions, changes at the periphery of the nucleus associated with ageing. The analyses showed that chromatin is compact and organised at the edges of the nuclei of young cells, while in aged cells it appears dispersed. As a proof of concept, the tool assesses how certain epigenetic drugs (source in Spanish) induce changes in chromatin organisation that are consistent with patterns more akin to those observed in young cells. This showcases the potential of the tool for screening treatments that could regenerate blood tissue and strengthen the immune system. The code and the image database are already available to the international scientific community. Collaboration with AI research teams As for the practical applications of this research, Dr Maria Carolina Florian, principal investigator at the Bellvitge Biomedical Research Institute (IDIBELL), explains to Euronews: "Currently, we have developed a tool that works as a proof of concept in murine haematopoietic stem cells. We are now working to optimise it and adapt it to human stem cells, with the hope of using it to evaluate new drugs that can act on nuclear architecture and be used to rejuvenate aged stem cells." Maria Carolina Florian tells Euronews that this work "has found the ideal ground to become a real tool thanks to the expertise of Dr Paula Petrone (BSC and ISGlobal). We plan to continue developing it and expand it for use with human stem cells. In addition, we are applying for further EU funding and seeking to establish links with other scientists in Europe who are interested in the practical application of our work." For her part, Paula Petrone believes that this research "shows that a deep-learning algorithm can tell apart young and aged stem cells by analysing images of cell nuclei stained with a low-cost substance called DAPI." 3D maps and open code to rejuvenate cellular architecture The internal map of the cell nucleus reveals a distinctive signature: in youth, chromatin is tightly entrenched at the edges of the nucleus; with age, that architecture falls apart. Identifying these patterns could help to study the decline of blood stem cells and to assess potential strategies capable of regenerating haematopoietic tissue and triggering changes throughout the body. To demonstrate the practical potential of the algorithm, the team designed a proof of concept: they applied the tool to aged cells exposed to various epigenetic drugs. The goal was to determine whether the molecules were able to produce changes in chromatin organisation consistent with a more youthful state. To accelerate new discoveries, the team has released the tool's code and a public database of 3D images of stem cells, with the aim of finding drugs that reinforce the immune system and quality of life in old age. Numerous international studies underpin the evidence for the fundamental role of the immune system in the ageing of the body, as Maria Carolina Florian points out: "In our laboratory, we have contributed by demonstrating the possibility of stimulating and rejuvenating aged haematopoietic stem cells to improve not only immune system function, but also life expectancy and the health of the organism as a whole." Long-term promise, caution for now: the science behind ChromAgeNet calls for prudence The new ChromAgeNet tool will help to understand the basic molecular mechanisms involved in the ageing of haematopoietic stem cells, which could become targets for therapies to rejuvenate them. "We hope to deepen our knowledge, consolidate our findings and identify new tools and drugs to improve the function of aged haematopoietic stem cells," Florian concludes. "In future, this technique could be used to screen new drugs with rejuvenating potential. By treating aged cells with different candidate compounds, AI could use image analysis to identify those that induce changes consistent with a more youthful state," Paula Petrone tells Euronews. For Dr Petrone, "Extending life expectancy is not as important as extending healthy life expectancy. The discovery of new drugs with rejuvenating potential could help, for example, to improve ageing of the immune system and thereby support healthier ageing." The researcher nevertheless stresses to Euronews the need for caution: "We must be very careful not to overestimate our results. These new technologies take years to be developed, validated and eventually reach the clinic. Our research is just one small contribution to the study of how we age and to the search for strategies that can support healthy ageing."