- SenSCOUT (Senescence Subtype Classifier Based on Observable Unique Traits) is an advanced tool developed to identify and classify distinct subtypes of cellular senescence using high-throughput single-cell imaging and machine learning. This innovative system was designed to address the growing recognition that cellular senescence is not a uniform state but a diverse and dynamic process.
- Rather than defining senescence through a limited set of markers, SenSCOUT captures the complex and subtle morphological and molecular differences among senescent cells, enabling a more precise and scalable approach to senescence profiling.
- The platform was initially applied to over 50,000 human dermal fibroblasts collected from donors aged 23 to 85 years. By analyzing various features—such as cell and nuclear morphology, texture, and the expression of known senescence-associated proteins—SenSCOUT was able to define 11 distinct cellular clusters. Among these, three subtypes—designated C7, C10, and C11—were characterized as bona fide senescent populations based on their expression of key senescence markers and lack of proliferation. Each subtype displayed unique morphological and molecular profiles, suggesting that senescence arises through multiple pathways and may serve different biological roles depending on context.
- One of the most striking findings was the identification of subtype C10 as strongly correlated with donor age, suggesting a possible link between this senescent phenotype and physiological aging. Furthermore, the different senescent subtypes responded variably to senolytic treatments, such as the combination of dasatinib and quercetin. Subtype C7, for instance, exhibited a higher sensitivity to senolytics, while others were more resistant. These results underscore the importance of senescence heterogeneity and suggest that personalized or subtype-specific interventions may be more effective for eliminating senescent cells in aging or disease.
- SenSCOUT’s integration of morphology-based classification with imputed protein expression via machine learning enables it to infer phenotypic states even in the absence of direct protein measurements. This makes the tool particularly useful for high-throughput screens and the analysis of archived imaging datasets. Moreover, its ability to classify senescence subtypes using non-destructive imaging techniques paves the way for potential applications in real-time, in situ monitoring of senescent cells in tissues or live-cell models.
