Gene Regulatory Network Reconfiguration in Direct Lineage Reprogramming

Abstract

In direct lineage reprogramming, transcription factor (TF) overexpression reconfigures Gene Regulatory Networks (GRNs) to convert cell identities between fully differentiated cell types. We previously developed CellOracle, a computational pipeline that integrates single-cell transcriptome and epigenome profiles to infer GRNs. CellOracle leverages these inferred GRNs to simulate gene expression changes in response to TF perturbation, enabling network re-configuration during reprogramming to be interrogated in silico. Here, we integrate CellOracle analysis with lineage tracing of fibroblast to induced endoderm progenitor (iEP) conversion, a prototypical direct lineage reprogramming paradigm. By linking early network state to reprogramming success or failure, we reveal distinct network configurations underlying different reprogramming outcomes. Using these network analyses and in silico simulation of TF perturbation, we identify new factors to coax cells into successfully converting cell identity, uncovering a central role for the AP-1 subunit Fos with the Hippo signaling effector, Yap1. Together, these results demonstrate the efficacy of CellOracle to infer and interpret cell-type-specific GRN configurations at high resolution, providing new mechanistic insights into the regulation and reprogramming of cell identity.

Defining Cardiac Recovery at Single Cell Resolution

“…To explore a relationship between RUNX1 in macrophages and fibroblasts in cardiac recovery, we applied a deep learning approach 55, 56 and found that RUNX1 target gene expression measured at the time of LVAD implantation predicts acquisition of recovered versus non-recovered states in both cell populations. We then used CellOracle 17 to build macrophage and fibroblast specific gene regulatory networks and ascertained the predicted effects of RUNX1 deletion using in silico transcription factor perturbation. Within macrophages, RUNX1 perturbation resulted in predicted loss of pro-inflammatory macrophages and pathogenic fibroblasts with shifts towards states observed in recovered hearts.…”

“…It is important to note that the in silico approach used to perturb Runx1 within macrophages and fibroblasts does not consider cellular crosstalk. While an important limitation, CellOracle transcription factor perturbation studies have been validated in vitro and in vivo and used by other groups 17, 29, 57 . We recognize that there is no perfect mouse model to recapitulate cardiac recovery 43 and the pressure overload JQ1 mediated recovery dataset has inherent limitations.…”

“…We then tested whether perturbation of RUNX1 could facilitate recovery in human macrophages and fibroblasts. We utilized CellOracle 17,29 to build cell specific gene regulatory networks and performed an in silico deletion of RUNX1. To define directional changes in cell fate resulting from RUNX1 perturbation, we constructed a vector (which was not certified by peer review) is the author/funder.…”

“…CellOracle (v0.10.5) 17 was used to perform GRN analysis in macrophages and fibroblasts separately. Processed data was imported into scanpy as .h5ad files.…”

“…We identified a RUNX1 gene regulatory network (GRN) in macrophages and fibroblasts that was associated with and predictive of cardiac recovery using a deep neural network. In silico perturbation analysis 17 , revealed that RUNX1 activity in macrophages and fibroblasts was predicted to control transition of cell states towards those associated with recovery and away from those associated with heart failure. Finally, we leveraged a mouse model of cardiac recovery 18 to validate the putative role of the Runx1 GRN in cardiac recovery.…”