Homesickness strikes Hepatocytes Deciphering the role of environmentally mediated deregulation of YAP/TAZ in dedifferentiation of in vitro cultured hepatocytes

This thesis manuscript consists of two separate and unrelated parts. In a first part, I will discuss the role of the Hippo pathway in stem cells and stem cell derived liver progeny. The second part, will be a treatise of a theoretical study of the cell cycle.

Part I

Today, primary human liver cells (especially hepatocytes, but also hepatic stellate cells) remain the gold standard in vitro model to study liver diseases and drug toxicity. These cells, however, are scarcely available and alternative cell sources are thus sought after. One such alternative are liver cells derived from human pluripotent stem cells (PSCs). Unfortunately, most currently available differentiation protocols for PSCs give rise to cells bearing an immature phenotype. Differentiation into hepatocytes, for example, results in so-called hepatocyte-like cells or HLCs, as they have a low expression of adult markers, lack major drug metabolizing enzymes (e.g. CYP3A4) and poorly repopulate the damaged liver of immunodeficient mice. PSC-derived stellate cells too are `immature' as they show an activated phenotype.

To address the maturation problem of PSC-derived hepatocytes, the Verfaillie lab has generated a PSC line that overexpresses three liver-specific transcription factors in a doxycycline inducible manner (termed HC3x cells) as well as an optimised differentiation medium. Differentiation of HC3x cells in the optimised medium resulted in a more mature hepatocyte progeny, with increased production of albumin and functional CYP3A4 in comparison to non-modified HLCs. In a first part of the results chapter, we describe how the more mature HC3x-HLCs fail to repopulate the liver of urokinase-type plasminogen activator (uPA)-SCID mice more efficiently than non-modified HLCs. These findings suggest that, even though HC3x-HLCs have improved drug metabolizing capacities as previously demonstrated by the lab, they still differ from primary hepatocytes (which do robustly repopulate murine livers).

It has been suggested that the discrepancies between HLCs and primary hepatocytes may (at least in part) emanate from the improper biomechanical features (cell-cell and cell-ECM interactions) of standard culture platforms. Although the underlying molecular mechanisms have not entirely been elucidated, indications exist that improper regulation of the Hippo pathway, which is a key player in mechanobiological signaling, impedes further maturation of HLCs. In a second part of the results chapter, we describe the analysis of RNAseq data that was available in the lab and show how the Hippo pathway is indeed improperly regulated in HLCs, and also in HC3x-HLCs, relative to unplated primary hepatocytes. Based on these findings, we exploited several strategies to manipulate the pathway in order to improve the maturation level of PSC-derived HLCs: small molecules, modification of the culture substrates, and genetically engineering the cells. Unfortunately, none of these strategies resulted in unambiguous conclusions or new insights, and future work will be required to further explore if and how manipulation of the Hippo pathway can close the maturation gap between HLCs and primary hepatocytes, and make them suitable for pharmaceutical studies and disease modeling or even regeneration of diseased liver in vivo.

The Hippo pathway does not only affect differentiation into hepatocytes, but has also been implicated in regulating pluripotency of stem cells and in the activation of hepatic stellate cells. In the last part of the results chapter, we discuss some findings related to the Hippo pathway in these cell types, and elaborate on why these experiments too could not be interpreted unambiguously, with a special focus on the use of doxycycline inducible overexpression systems.

Part II

Amplification of genomic material and its segregation among daughter cells are fundamental for the proper development, growth and reproduction of a living organism. These two processes are part of a larger series of periodically ordered events, commonly referred to as the cell cycle. The dynamical features of the cell cycle, one of which is bistability, emanate from complex protein interaction networks at the molecular level. Indeed, several feedback loops have been identified (both theoretically and experimentally) that give rise to bistable behavior of cyclin dependent kinases (CDKs), which coordinate proper cell cycle progression. More recently, also the CDK-counteracting phosphatase PP2A-B55 has been proposed to act in a bistable manner, but how exactly this bistability is established remains an open question, as at least two distinct mechanisms have been proposed in the literature. Here, we used a mass-action model of the core biochemical interactions regulating the activity of PP2A-B55 to compare two potential mechanisms generating bistability. Based on this theoretical model, we were able to propose perturbations that can be exploited to distinguish between both options in future experiments.

Mechanistic mass-action models containing molecular details often result in a large number of variables and parameters. As this might complicate the interpretation and generalization of the obtained results, it is often desirable to reduce the complexity of the model. One way to accomplish this is by replacing the detailed reaction mechanisms of certain modules in the model by a mathematical expression that qualitatively describes the dynamical behavior of these modules. Such a phenomenological approach has been widely adopted for ultrasensitive responses, for which underlying reaction mechanisms are often replaced by a single Hill function. In contrast, however, S-shaped response curves, which are often encountered in bistable systems, are not easily modeled in such an explicit way. In a second part of the results chapter, we describe how the classical Hill function can be extended into a mathematical expression for an S-shaped response. Using the early embryonic cell cycle as our example, we show how this approach greatly reduces the complexity of the mathematical model, while it faithfully reproduces the results of the mass-action model described in the first part of the chapter. Lastly, we show how the phenomenological model can be exploited to model the overall cell cycle as a chain of interlinked bistable switches, and can incorporate events such as the restriction point or DNA damage.