During cell differentiation, an immature unspecialized cell assumes the stable and lasting phenotype of a specialized cell type. Although this process is often considered to be deterministic and regulated by instructive signals, the stochastic nature of cell fate determination has long been recognized. In fact, cells within a clonal population exposed to the very same environment can exhibit different phenotypes. Such "non-genetic heterogeneity" may either be due to "gene expression noise" or to slower fluctuations of protein levels, implying transient cell-individuality. However, whether such cell-to-cell variability may account for the stochasticity of cell fate decision in mammalian cells remains unknown.

In this dissertation, I examine the role of non-genetic heterogeneity in mammalian cell fate determination. Using human promyelocytic precursor HL60 cells, I first demonstrate that cell differentiation is a multi-step, switch-like process at the individual-cell level. In view of such discrete transitions, non-genetic cell heterogeneity becomes biologically important, which I thus investigated closer. Within clonal populations of murine hematopoietic progenitor EML cells, "outlier" cells with extreme expression levels of the stem cell marker Sca-1 reconstituted the parental Sca-1 distribution with surprisingly slow kinetics. The cells with extreme high and low Sca-1 also differed in their preference for commitment to the erythroid or myeloid lineage. This difference was reflected in their transcriptomes, which included dramatic differences in basal levels of fate-determining transcription factors. This spontaneous variability in cell-fate priming naturally resolves the old dualism between the "selective" and "instructive" models of cell fate determination, since it provides the variation that is inherently required for selection, allowing differentiation signals to selectively instruct only the responsive subset of cells and influence their gene expression. This insight could be utilized to increase efficiency in attempts to steer stem cell differentiation to a desired fate. Finally, I constructed a mammalian cell system for simultaneous measurement of expression of the lineage-determining transcription factor PU.1 and of its downstream target Mac-1 by fluorescence microscopy. This experimental cell system will enable us to track the origin and propagation of gene expression fluctuations in single, live hematopoietic progenitor cells just undergoing differentiation.