(545c) Computational Model of Single Cell Transcriptional Regulation and Cellular Networks Driving Liver Regeneration Following Surgical Resection

Liver resection is used clinically as a treatment for hepatocellular carcinoma and as a technique to enable live liver transplant. In both cases, the resected liver initiates a highly orchestrated regeneration program that is expected to culminate in complete restoration of liver mass. Despite decades of study, our understanding of the regulatory mechanisms underlying this repair process remains incomplete. This lack of complete understanding limits the ability to intervene in cases where a resected liver fails to regenerate to the levels required to support normal physiological function. Nevertheless, the availability of clinical data and pathology samples from regenerating and non-regenerating cases present an opportunity to employ modeling approaches that integrate data from different functional scales to better understand the mechanisms underlying clinical resection responses. Our present study seeks to address the gap in mechanistic understanding of liver regeneration by using a combination of single-cell based transcriptional analysis and computational network modeling, which can then be integrated with clinical data to improve prognostic outcomes.

Initiation and progression of regeneration relies on the coordination of the responses of multiple cell types over time. A canonical description of the regeneration process after resection includes transient activation of the resident macrophages by the injury, followed by priming of hepatocytes to re-enter cell cycle, growth factor production by hepatic stellate cells and sinusoidal endothelial cells, proliferation of hepatocytes, proliferation of non-parenchymal cells, and termination of proliferative processes via remodeling the matrix microenvironment. Disruption of this coordinated sequence can lead to deficits in liver mass recovery, as clinically observed in chronic liver disease.

We developed a computational model that accounts for the multi-scale nature of liver regeneration by integrating physiological-scale interactions; activation phenotypes of hepatocytes and liver non-parenchymal cells; and molecular signaling networks. We explored a range of model dynamics to identify parameter sets that account for experimentally observed regeneration profiles of multiple animal models of liver disease as well as the human liver transplant case. Model simulations predicted that aberrant activation dynamics of hepatocytes and hepatic stellate cells are key factors suppressing liver regeneration in multiple disease contexts.

We tested these predictions in the alcoholic liver disease case by acquiring a new experimental data set on gene regulation in hundreds of single and pooled cells obtained through laser capture microdissection of liver tissue. We analyzed the high-dimensional single cell scale gene regulation data to identify distinct cell populations, and extrapolated from the single-cell scale to cell phenotype distribution in whole tissue. Our approach characterized the transcriptional state of individual cells in the in vivo regenerating liver at a level of detail not previously achievable. Characterization of transcriptional profiles revealed cell-type specific, functionally distinct states for hepatocytes as well as stellate cells in our data. Surprisingly, we found the two liver cell types distributed among all identified functional states irrespective of the experimental condition (baseline or regenerating, alcohol adapted or control). However, relative proportions of the cells distributed among the identified functional states changed based on regeneration stage as well as adaptation to chronic ethanol intake, consistent with predictions of the computational model. These results suggest that chronic alcohol use impairs liver regeneration via significant alteration of cellular networks that are critical to initiation and progression of liver repair following resection. Our integrated experimental and computational analysis points towards new avenues for therapeutic intervention based on renormalizing the cellular functional states for improving surgical outcomes.