Random versus deterministic input into stem cell lineage choice

Research
|

Random versus deterministic input into stem cell lineage choice

"Everything ... depends on these and only on these: A set of fixed, deterministic laws. A purely random set of accidents." -- Marvin Minsky

One of the hallmarks of stem cells is their multipotentiality. However, despite this being a key feature of their "stemness", we are still working to achieve a conceptual understanding of the nature of this potential. While there have been several studies done thoroughly identifying and compiling lists of genes expressed in stem cells, the relative contributions of different forces and factors to their lineage choice remains an active area of investigation. A better understanding of lineage specification will help us more optimally use stem cells therapeutically, as well as tools in large scale screens by increasing the efficiency with which we can coax stem cells toward a particular cell type.

Bipotent progenitor cells provide an ideal system in which to gain a better conceptual understanding of cell lineage specification. One of the major questions in this area concerns whether cells differentiate toward a particular fate as a result of intrinsic or extrinsic signals, two not mutually exclusive models. An instrinsic model relies upon stochastic processes such that expression levels of the same gene in progenitor cells vary as the cells move through metastable states toward a more stable "attractant" state. In the intrinsic model, cells are responding to a preexisting program acquired by chance and external factors such as cytokines and growth factors act merely to promote cell viability, rather than play an instructive role (Huang et al., 2007). In contrast, a deterministic, or externally prompted control of lineage choice invokes external factors playing an active and instructive role in lineage determination. In an extrinsic model, external factors control cell fate by imposing the activation or repression of particular genetic networks.

Perhaps a more biologically relevant model is one that integrates the contributions of both stochastic and determinate modes of differentiation. Such a unified model has been proposed and is described by a progenitor cell that may express genes associated with the determination of many lineages, and this low-level background expression may provide the groundwork for amplification or repression of particular lineage determinants as a result of stochastic changes in necessary regulatory components, further exacerbated by external factors (Envers et al., 2008). Recent work exploring the detailed dynamics of different steps in the lineage determination process to better understand the relative contribution of random gene expression levels versus external deterministic factors supports a model that incorporates both intrinsic and extrinsic components and indicates that the progenitor cell state is primed by a gene expression profile associated with multiple lineages.

Sui Huang from the Harvard Stem Cell Institute and colleagues investigated how different ratios of lineage-determining transcription factors in the cell of bone marrow-derived bi-potential progenitor cells contribute to cell fate specification (Huang et al., 2007). Huang and colleagues examined the well-characterized blood cell differentiation system. The common myeloid precursor cell further differentiates to adopt either an erythroid/megakaryocyte or myelomonocytic lineage. Two transcription factors, GATA1 and PU.1 play key roles in this decision. While these factors are known to be important, the question of how the bipotent state is maintained prior to lineage commitment and what eventually prompts a cell to make such a choice is not clearly understood. GATA and PU.1 have been shown to directly regulate networks of different lineage specific genes. They repress each other and can also autoregulate their own transcription. The researchers had previously proposed that autoregulation of GATA and PU.1 promoted and stabilized the progenitor state (so-called "multilineage priming") and it was the balance between auto and cross regulation that promoted differentiation. Following up on these ideas, the researchers created a mathematical model based on the GATA1 and PU.1 system to recapitulate binary cell fate decision and compare their results with gene expression data. From these studies, the researchers propose that commitment to a lineage occurs in two stages. First, the progenitor state, which represents a metastable state, is destabilized and then second, the cell is forced to choose between alternative states representing distinct lineage commitments in order to reach another stable state. This model has components of both the stochastic and instructive/deterministic models of differentiation and thus may represent a paradigm for other systems.

While the proposed model describes progenitor status and lineage choice in terms of metastable states, further experiments describe the genetic read out of these different states. Within a clonal population of cells there are heterogeneous phenotypes that may be accounted for by variations in protein levels. In order to understand if these fluctuations in protein levels contribute to stochastic differentiation of stem cells, Chang et al. (2007) characterized clonal populations of hematopoietic progenitor cells (HPCs). The researchers assayed multipotent mouse HPCs for protein levels of the stem cell marker Sca-1. Within a clonal cell population they sorted cells into groups with high, medium, and low Sca-1 expression. These sorted fractions individually repopulated and the researchers found that each of the groups did so slowly recapitulating the heterogeneous "bell-shaped" histogram mix of expression levels they saw in the original clonal population. The researchers wanted to then understand what drives the consistent regeneration of this bell shaped curve from sorted fractions. Since the genetics of Sca-1 regulation is not well understood, the group explored a number of theoretical frameworks for stochastic processes to explain the results. From these studies, the researchers concluded that the different levels of Sca-1 expression reflect distinct metastable states of the cell and that this range of metastable states helps to prime the cells toward lineage commitment. These findings are highly significant in a greater biological context because it is often assumed that lineage adoption is achieved by a group of mainly homogeneous cells and those with drastically different gene expression patterns represent outliers. Instead, the heterogeneous distribution of expression levels is critical to priming cell fate decisions.

A better understanding of the relative stochastic and instructive nature of cell lineage choice and the nature of multipotentiality has significant implications for stem cell based therapeutics and for optimizing stem cells as tools for large scale screens. The heterogeneity within a clonal population of cells represents fractions with varying predilections toward targeted differentiation to a particular cell type. This may explain the extremely low efficiency of targeted differentiation many researchers encounter. By characterizing these predilections, we may soon be able to work with the fraction of stem cells that are more likely to respond to a particular differentiation protocol, thus dramatically increasing the robustness of the protocol and defining a more reliable cell source.

References

  • Chang, H.H., Hemberg, M., Barahona, M., Ingber, D.E., Huang, S. (2008). Transcriptome-wide noise controls lineage choice in mammalian progenitor cells. Nature 453, 544-7.
  • Enver, T., Heyworth, C.M., Dexter, T.M. (1998). Do stem cells play dice? Blood 92, 348-51.
  • Huang, S., Guo, Y.P., May, G., Enver, T. (2007) Bifurcation dynamics in lineage-commitment in bipotent progenitor cells. Dev Biol. 305, 695-713.