Stem cells are characterized by their unique ability to generate both self-renewing and differentiating daughter cells. Our group addresses the molecular mechanisms that establish these different properties within stem cell lineages. For this, we use both Drosophila and mouse genetics and focus on the developing brain where these mechanisms are essential to regulate neurogenesis and prevent tumor formation.
In the Drosophila brain, neural stem cells called neuroblasts undergo repeated rounds of asymmetric cell division (Fig. 1A). One of the resulting daughter cells continues to divide in a stem cell-like manner while the other cell terminally divides into two differentiating neurons. During each neuroblast division, the cell fate determinants Numb, Prospero and Brat segregate into the smaller, basal daughter cell where they prevent self-renewal and induce differentiation (Fig. 1A, B). This happens, because the protein kinase aPKC localizes to the opposite, apical side and removes the determinants by phosphorylating their membrane localization domains. At the same time, aPKC associates with microtubule binding proteins to ensure that the mitotic spindle is set up in an apical-basal orientation. As a result, the determinants are specifically inherited by the basal daughter cell. In the absence of Brat, Numb or Prospero, both daughter cells retain the ability to self-renew. As a consequence, stem cells expand exponentially and overgrow the brain. Eventually, they form gigantic brain tumors that can be propagated by serial transplantation into host flies, where they become aneuploid and undergo metastasis (Fig. 1C). Understanding, how the transcriptional changes induced in one daughter cell irreversibly target this cell to differentiation and how defects in this reprogramming event lead to tumorigenesis is one of our main goals.
We have used transgenic RNAi to identify over 600 genes that regulate self-renewal in Drosophila neuroblasts. Among those are 18 tumor suppressors that cause neuroblast overproliferation. Besides the asymmetric cell division machinery, this set includes six nuclear proteins that we believe form the transcriptional machinery acting downstream of the segregating determinants. Three of these are part of the SWI/SNF chromatin-remodeling complex, one is a known binding partner of Histone deacetylase and two are implicated in the control of transcriptional elongation. We have recently established a technology that allows us to purify neuroblasts and their differentiating daughter cells in large quantities and to determine their transcriptomes by deep sequencing technology (Fig. 2A). Together with the enormously powerful genetic tools available in Drosophila, this allows us to determine the transcriptional changes upon removal of any of the nuclear regulators in a time resolved manner. We have been able to establish a network of transcription factors that act in neuroblasts and establish a stable self-renewing state (Fig. 2B). Development of those resources will hopefully soon allow us to understanding, how the transcriptional network is irreversibly changed after asymmetric cell division in the differentiating daughter cell and how defects in asymmetric cell division drive the cells towards tumor formation.
In the mouse brain, progenitor cells called radial glia generate neurons of the cortex through lineages that are strikingly similar to Drosophila neuroblasts. Initially, progenitors expand through symmetric divisions but later, they divide asymmetrically giving rise to differentiating daughter cells as well (Fig. 3). In contrast to flies, however, the mechanisms that establish this asymmetry in mice are largely unknown. We are using our knowledge from Drosophila to understand, how those asymmetric divisions are regulated.
The machinery for asymmetric cell division is conserved between flies and mice. To test its role, we mutated the gene inscuteable, a specific regulator of asymmetric cell division and spindle orientation in Drosophila. In neuroblasts, inscuteable is essential for aPKC to orient the mitotic spindle. In mice, inscuteable is required for spindle orientation as well. In inscuteable knock-out mice, the characteristic re-orientation of cell division that is observed when progenitors switch from symmetric to asymmetric division (Fig. 3B,C) is not observed. Instead, progenitors continue to divide parallel to the surface even late in neurogenesis. As a consequence, lineages shift from indirect to direct neurogenesis, generating neurons instead of intermediate progenitors. Therefore, inscuteable mutant mice have less cortical neurons while inscuteable overexpression has the opposite effect. These results shed light on the role of spindle orientation during mammalian development and provide a surprising answer to a long standing question in the field of mammalian development. As the expansion of brain size from mice to humans involves intermediate progenitors that divide even more than once, inscuteable might play a key role in the evolution of the mammalian neocortex. We are currently extending our experiments to human model systems to test this hypothesis and to analyze the connections between spindle orientation and brain developmental disorders.