Periodic Reporting for period 1 - DesignerCells (Building a causal regulatory network of mesenchymal cells using targeted programming experiments)
Reporting period: 2021-09-01 to 2023-08-31
We addressed the problem of how transcription factors, the central proteins involved in gene regulation, shape cell identity when they are activated in a stem cell. To do this, we overexpressed the transcription factors and analyzed single-cell gene expression data along with the dosage that each cell received of the transcription factor. This is important for society because transcription factor overexpression is a main way in which cells can be manipulated to develop new cell therapies. Furthermore, most diseases involve some dysregulation in transcription factors, and as such understanding this is important to understand the mode-of-action of existing and future drugs. The overall objective is to assess how various cellular processes, such as differentiation and cell proliferation, interact with the dose of transcription factors, and how cellular heterogeneity can be formed.
Overall, we identified extensive interactions between various transcription factors and these cellular processes. For example, we found that nearly every TF interacts strongly with the cell cycle. Extensive dosage-dependencies also exist for many transcription factors.
Overall, we identified extensive interactions between various transcription factors and these cellular processes. For example, we found that nearly every TF interacts strongly with the cell cycle. Extensive dosage-dependencies also exist for many transcription factors.
I helped with the experimental design and execution of the generation of a new single-cell transcriptomics dataset containing 6 transcription factors. This dataset contained a high number of cells (>200) for each transcription factor, which enabled us to map the fine changes of dosage sensitivity genome-wide across varied levels of dosage. We also included several combinations of transcription factors to be able to learn how two potentially synergistic or antagonistic transcription factors would interact. I created a computational method that can enable us to accurately assess and iteratively refine different potential biological mechanisms that may be underlying the dataset at hand. The various common cellular processes, modalities, and mechanisms (Figure 1c) that are used to understand transcription factor data are included, but it is easy to expand with other processes/modalities depending on the biological system. This removes the monolithic nature of previous frameworks and democratizes model building. We found that dosage and cell cycle have much more extensive effects than previously thought. We found extensive number of non-monotonic gene responses, particularly in Homeobox TFs. Gradients of homeobox TF expression have been described in several studies, often caused by upstream gradients in morphogen concentration during development. Our analysis suggests that non-monotonic transcriptional regulation is a general, feature of this TF family and that these TFs often alter their mode of gene regulation depending on their concentration. We also applied Latenta to understand how the cell cycle may interact with, and here again found that cell cycle interactions are the rule rather than exception. We also observed considerable differences in dosage dependencies when analyzing combinations of TFs, one of the key objectives within the DoA. In particular, we found that the antagonism or synergism between two TFs can change depending on the respective dosages of the TF, with a second TF being seemingly synergistic if its dosage is relatively low compared to the main TF under study.
I was fortunate to receive positive feedback for our projects by presenting it at meetings at companies (Nestle) and where people from companies were present (Single-cell genomics Hinxton conference, with people from Jansen and Roche). Indeed, our computational toolkit ChromatinHD has been accessed and installed by over 180 users (between its publication of 24/07/2023 until 23/08/2023), and I have received feedback from several users from various institutes (MIT, Caltech) through GitHub issues.