Final Report Summary - THSWITCH (Decoding genetic switches in T helper cell differentiation)
The T helper (Th) cell system is part of the adaptive immune system and has an important role in instructing other cells of the immune system to mount immune responses. Th cells are characterized by the surface expression of the CD4 coreceptor. Upon binding of antigen by the T cell receptor (TCR), naïve T helper cells become activated and undergo several rounds of cell division, if one or more cytokines are present. Alongside this proliferation process, the cells differentiate towards one of several mature Th subtypes. In addition to T regulatory cells (Tregs), which fulfill immunosuppressive roles, three main effector subtypes are known, Th1, Th2 and Th17, all displaying a characteristic influence upon the outcome of an immune response by secreting a small number of ‘signature’ cytokines. This process is important not only for combating infections, but also cancer, and imbalances are directly responsible for autommunity and allergies.
Besides these applications, the T helper cell system is a good model for a more basic study of switches in gene expression during differentiation for a number of reasons, both conceptual and practical. At the conceptual level, this system consists of a precursor (naïve Th) with different possible differentiation outcomes. Certain combinations of differentiated cells are known to be able to interconvert, providing a model of cellular plasticity. Therefore, there are several different switches between cell types that can be studied using this system. The cell types are uniquely identified through their expression of different cytokines, so these molecules provide a direct, quantitative readout of the phenotypic state of a cell. We profiled six different in vivo and in vitro cell states using RNA-sequencing to show the cytokine, cell surface receptor and transcriptional regulatory signatures of these states. We provide this data online at Th-express.org.
The switches between these cell types are likely to have statistical properties that involve extensive stochasticity, meaning that there are probabilities associated with differentiating, interconverting etc depending on the particular cellular context. We describe this in a review (Hebenstreit et al, Curr Op Cell Biol, 2012), and have shown that activated cells proliferate twice as fast if they are cytokine-secreting (Proserpio et al., submitted).
During the past five years, we have followed several lines of investigation. First, we developed that technical skills to carry out high-throughput single cell transcriptomics in order to be able to profile cells genome-wide at the resolution of an individual cell. This is particularly valuable when cell samples are as heterogeneous as T helper cells, because it allows one to capture the identity of one cell at a time. We were the first group in Europe to publish this using a microfluidics robot (Fluidigm C1), and showed how to distinguish technical from biological variation using synthetic standards for calibration (Brennecke et al., Nature Methods, 2013).
Secondly, we developed many statistical and computational ways of analyzing this type of data, which we have made publicly available in the github online repository. These computational methods include a way of accounting for technical noise (Buettner et al., Nature Biotech, 2015) and a way of assigning the cell cycle stage of each cell (Scialdone et al. Methods, 2015). Importantly, we have made a software pipeline to identify which T cells belong to the same clone or “family” of cells by reconstructing the T cell receptor sequences from single cell RNA-sequencing data. This allows us to trace the lineages of T cells as they develop and expand during an immune response. We call this pipeline TraCeR (Stubbingbton et al., BioRxiv, 2015; Stubbington et al., Nature Methods, 2016, in press).
Finally, we have used our innovative approaches to gain fundamental new biological insights. The TraCeR software pipeline allowed us to show for the first time in a living organism that T cells from the same clonotype exist in distinct cell states (Stubbingbton et al., BioRxiv, 2015; Stubbington et al., Nature Methods, 2016, in press). Earlier, we showed that T cells not only use cytokines to signal, but may have a previously unappreciated signaling system based on steroids (Mahata et al., Cell Reports, 2014).
Besides these applications, the T helper cell system is a good model for a more basic study of switches in gene expression during differentiation for a number of reasons, both conceptual and practical. At the conceptual level, this system consists of a precursor (naïve Th) with different possible differentiation outcomes. Certain combinations of differentiated cells are known to be able to interconvert, providing a model of cellular plasticity. Therefore, there are several different switches between cell types that can be studied using this system. The cell types are uniquely identified through their expression of different cytokines, so these molecules provide a direct, quantitative readout of the phenotypic state of a cell. We profiled six different in vivo and in vitro cell states using RNA-sequencing to show the cytokine, cell surface receptor and transcriptional regulatory signatures of these states. We provide this data online at Th-express.org.
The switches between these cell types are likely to have statistical properties that involve extensive stochasticity, meaning that there are probabilities associated with differentiating, interconverting etc depending on the particular cellular context. We describe this in a review (Hebenstreit et al, Curr Op Cell Biol, 2012), and have shown that activated cells proliferate twice as fast if they are cytokine-secreting (Proserpio et al., submitted).
During the past five years, we have followed several lines of investigation. First, we developed that technical skills to carry out high-throughput single cell transcriptomics in order to be able to profile cells genome-wide at the resolution of an individual cell. This is particularly valuable when cell samples are as heterogeneous as T helper cells, because it allows one to capture the identity of one cell at a time. We were the first group in Europe to publish this using a microfluidics robot (Fluidigm C1), and showed how to distinguish technical from biological variation using synthetic standards for calibration (Brennecke et al., Nature Methods, 2013).
Secondly, we developed many statistical and computational ways of analyzing this type of data, which we have made publicly available in the github online repository. These computational methods include a way of accounting for technical noise (Buettner et al., Nature Biotech, 2015) and a way of assigning the cell cycle stage of each cell (Scialdone et al. Methods, 2015). Importantly, we have made a software pipeline to identify which T cells belong to the same clone or “family” of cells by reconstructing the T cell receptor sequences from single cell RNA-sequencing data. This allows us to trace the lineages of T cells as they develop and expand during an immune response. We call this pipeline TraCeR (Stubbingbton et al., BioRxiv, 2015; Stubbington et al., Nature Methods, 2016, in press).
Finally, we have used our innovative approaches to gain fundamental new biological insights. The TraCeR software pipeline allowed us to show for the first time in a living organism that T cells from the same clonotype exist in distinct cell states (Stubbingbton et al., BioRxiv, 2015; Stubbington et al., Nature Methods, 2016, in press). Earlier, we showed that T cells not only use cytokines to signal, but may have a previously unappreciated signaling system based on steroids (Mahata et al., Cell Reports, 2014).