Reversing pathological cell states: the road less travelled can extend the therapeutic horizon

Abstract

Acquisition of omics data advances at a formidable pace. Yet, our ability to utilize these data to control cell phenotypes and design interventions that reverse pathological states lags behind. Here, we posit that cell states are determined by core networks that control cell-wide networks. To steer cell fate decisions, core networks connecting genotype to phenotype must be reconstructed and understood. A recent method, cell state transition assessment and regulation (cSTAR), applies perturbation biology to quantify causal connections and mechanistically models how core networks influence cell phenotypes. cSTAR models are akin to digital cell twins enabling us to purposefully convert pathological states back to physiologically normal states. While this capability has a range of applications, here we discuss reverting oncogenic transformation.

Keywords: cell state transition assessment and regulation method; control of cell state transitions and fate decisions; digital cell twins; omics data.

Figure 1.. cSTAR pipeline help us deliberately convert cell states.

cSTAR uses omics data as the input and ML to classify distinct cell state and phenotypes. It constructs the STV in the molecular feature space that indicates a path between distinct cell states, followed by the selection of high ranked STV contributors, which determine components of a core network controlling state transitions. Systematic perturbations and omics data on perturbation responses are used to infer directions and strengths of causal connections of the core network (including feedback loops) by BMRA. Incorporation of the DPD module summarizes cell-wide network and links molecular features to the cell phenotypes. Network reconstruction is followed by a mechanistic model of the core network and cell state transitions. This model predicts cell responses to small molecule therapeutics and transitions between cell states in Waddington’s landscape, which must be validated experimentally.

Figure 2.. Signaling networks shape the landscape of cell fate decisions.

A now famous model of Waddington’s landscape describes cell state changes by cell’s movement in a landscape of mountains and valleys, in which different valley corresponds to different cell fate decisions [76]. Pseudotime methods allow identifying cell’s trajectory at the landscape, and RNA velocity methods allow reconstructing the landscape in the transcriptomics space. The advantage of cSTAR is that it reconstructs both cell’s trajectory and Waddington’s landscape, but also connects it to the dynamics of signaling networks (shown as connected nodes shaping cell fate decisions). Not only cSTAR helps us understand cell state transitions but allows to steer this process in desired direction.

Figure I Box I.. A diagram of BMRA pipeline.

BMRA, Bayesian MRA formulation, uses the two following inputs: (1) measured responses to perturbations and (2) a prior network, derived from the existing knowledge if it is available, or a non-informative prior network (generated by the equal probability distribution for the presence or absence of a connection when the prior knowledge does not exist). In contrast to the deterministic MRA, BMRA allows to reconstruct networks using the datasets in which not all nodes are directly perturbed. MRA equations serve as the likelihood function for BMRA. Prior network connections and experimental data are fed to the Bayes rule to generate a posterior distribution of network connections and their strengths.