Bombarded daily with threats both external and internal, our immune system has the remarkable ability to adapt to virtually any situation thanks to its diverse array of cells. However, certain conditions (e.g. autoimmune disease, hypo-responsiveness) result from an imbalance of the types and reactivity of immune cells. Being able to reliably identify which immune cell subsets are involved in disease could facilitate the development of better diagnostic tools and treatments.
“Most of the time, the biomarkers that arise from blood gene expression data are not interpreted from an immunological point of view, even though the cells being interrogated are immune cells,” explained Anis Larbi, a Principal Investigator at the Singapore Immunology Network (SIgN), A*STAR. To enable researchers to link gene expression data to specific cell types, Larbi and his team have developed a method to identify different immune cell subsets in patient blood samples.
Using a combination of high dimension flow cytometry, fluorescence-assisted cell sorting, RNA sequencing and microarray data, the researchers were able to definitively characterize 29 different immune cell types. Beyond simply assigning a molecular ‘fingerprint’ to immune cell types, the researchers were also able to infer the proportions of each immune cell population by using a statistical model known as the ‘robust LM’ method, which allowed them to obtain absolute gene expression values.
“By adding a deconvolution tool to other existing tools to understand gene expression data, one can now have a very precise idea of which cells were present in the studied samples, without the need to perform additional analysis such as high dimension flow cytometry, which requires special skills and equipment,” Larbi explained.
The researchers demonstrated the power of their method by using it to examine the immune response of individuals after influenza vaccination, showing how the composition of immune cell populations shifted in response to vaccination.
These findings have gone towards the development of an app for use by the broader scientific community. “This work should enable us to predict immune cell subset frequencies from previous studies where only gene expression data were available,” Larbi pointed out.
Moving forward, the group is planning on using single-cell data for the deconvolution of more cell subsets than ever performed. Hinting at potential clinical applications, Larbi suggested, “Studies focusing on immune cell composition in situ will help us accelerate the progress in the field of cancer immunotherapy.”
The A*STAR-affiliated researchers contributing to this research are from the Singapore Immunology Network (SIgN).