The mechanisms that regulate INM in apical progenitors are poorly understood. For example, nuclei migrate apically during G2 phase and basally during G1 phase, but whether these movements occur via microtubules or actin is controversial. It has been suggested that mitosis initiates at the apical surface because the centrosome is positioned there, but initiation of mitosis has not yet been observed in live tissue through time-lapse microscopy.
Apical nuclear movement
Nuclei migrate toward the apical surface during G2 phase. There has been disagreement about the specific mechanisms that regulate apical nuclear migration. Microtubules and the minus-end-directed microtubule motor protein, dynein, are believed to be required for apical nuclear migration in the cerebral cortex (Tsai et al. 2005, 2007) because a considerable number of dynein or microtubule-associated proteins, when perturbed, disrupt INM there: (i) Lis1 is a regulator of dynein. RNAi of Lis1 in the mouse cerebral cortex prevents apical nuclear migration and inhibits mitosis, suggesting that dynein moves nuclei apically along microtubules (Tsai et al. 2005). Also, a reduction in Lis1 causes ectopic mitosis in apical progenitors of the cerebral cortex (Gambello et al. 2003). (ii) NudC, another regulator of dynein, is also required for apical nuclear movement in the rat neocortex (Cappello et al. 2011). (iii) Cep120, a centrosomal protein, and (iv) transforming acidic coiled-coil proteins (TACCs) associated with the centrosomes are necessary for the integrity of the microtubule cytoskeleton. These proteins are also essential for INM in the mouse neocortex (Xie et al. 2007). (v) RNAi of centrosomal proteins Hook3 or PCM1 results in a reduction of pericentriolar satellites, reduces microtubule anchorage at the centrosomes, and impairs INM in the mouse neocortex (Ge et al. 2010). (vi) TPX2 appears to organize the microtubule array in G2 phase and is also essential for normal nuclear movement during INM (Kosodo et al. 2011). (vii) Decreased levels of dynactin, an activator of dynein, result in increased average nuclear distance from the apical surface during interphase and slower apical nuclear migration in zebrafish retinal progenitors (Del Bene et al. 2008). (viii) Inhibition of casein kinase 2 in the rat retina causes a loss of microtubule integrity and slows INM (Carneiro et al. 2008). (ix) The KASH protein Syne-2 and the SUN proteins, SUN 1 and SUN 2, are important for INM in the mouse retina (Yu et al. 2011). KASH and SUN proteins bridge the nuclear envelope and are involved in many nuclear migration events (reviewed Starr 2009). Syne 2 interacts with kinesin and dynein in mouse brain and retinal lysates, and colocalizes in immunohistochemistry with dynein and kinesin in mouse retinal progenitor cells and mouse cerebral cortex apical progenitors (Zhang et al. 2009; Yu et al. 2011). Taken together, these studies are consistent with a model of apical nuclear migration where nuclei migrate along microtubules via dynein toward the apical centrosome.
Evidence also suggests that actin and myosin are involved in apical nuclear migration during INM. Cytochalasin B, an actin depolymerizing agent, has long been known to disrupt INM and cause ectopic mitotic figures in the neural tube. The interpretation has been that the drug inhibited apical nuclear migration and that mitosis occurred wherever the nuclei stopped. This was never shown with time-lapse imaging, however (Karfunkel 1972; Murciano et al. 2002). In the zebrafish retina, myosin inhibition with blebbistatin also stops apical nuclear migration, suggesting that actomyosin drives the process. Consistent with a critical role for actomyosin in INM, Norden et al. (2009) reported that colcemide-induced depolymerization of microtubules, and loss of dynactin function with a dominant-negative p150 construct does not affect apical nuclear movement in zebrafish retina.
Actin-driven and microtubule-driven mechanisms are not necessarily mutually exclusive. For example, both actomyosin and microtubules are required for cell division and maintenance of epithelial structure (Hildebrand 2005). Results from our laboratory unexpectedly revealed that apical movement during INM is a two-step process. Nuclei move apically during G2 phase, but do not reach the lumen of the chick neural tube during interphase. Instead, nuclei enter M phase marked by nuclear envelope breakdown (NEBD) on average 16 μm away from the apical surface. The mitotic cell then rounds up the rest of the way to the apical surface. We find that nuclear movement during G2 phase is dependent on microtubules, whereas apical rounding during mitosis is dependent on actin. Disruption of the microtubule cytoskeleton with colcemide prohibits apical nuclear movement during G2 phase, but neither prevents mitosis, nor stops cells from moving to the apical surface during mitosis. Conversely, cytochalasin B does not stop intact nuclei from moving apically during INM, but instead causes ectopic divisions by preventing mitotic cells from rounding up to the apical surface. Thus, the seemingly contradictory results could be explained by two parts of the process being controlled by microtubules and actomyosin.
It is also possible that the disagreement over the mechanism comes from the relative size of the tissues being studied and markers being used. By some estimates, the zebrafish retina is three to four nuclei thick in the apical-to-basal direction, whereas the mammalian cortex is closer to 10 nuclei in width (Miyata 2007; Taverna & Huttner 2010). Different mechanisms may be required to move the nuclei the greater distances observed in thicker regions of the neuroepithelium. FC Sauer noted a difference in nuclear movement between the neural tube and the thinner garter snake epithelium of the gut. In the neural tube, nuclear size (as a marker for stage in the cell cycle) and position were consistent with nuclei moving gradually during G1 and S-phase, as the nucleus gets larger. In the gut epithelium, size and position are not as well correlated, suggesting that mostly stochastic movements occur during interphase (Sauer 1936). Similarly, Norden et al. (2009) report that basal nuclear movement in the shorter zebrafish retina and hindbrain is also less gradual and more punctuated than published observations in mammalian cerebral cortex, taking place only early in G1 phase (Tsai et al. 2005; Leung et al. 2011). We propose that microtubules and microtubule motor proteins may be required for the longer, more gradual nuclear movements observed during G1 and G2 in the elongated apical progenitor cells of the cerebral cortex. Microtubules may be insufficient to move the nucleus all the way to the apical surface, however. Myosin and actin may be sufficient to move the nucleus in a more stochastic manner over a more limited range observed in shorter epithelial cells, and may be required for ensuring mitotic cells are completely rounded up to the apical surface (Fig. 1).
Figure 1. Different lengths of epithelial cells may dictate the mechanism of apical nuclear movement for mitosis. Microtubule-dependent and actin-dependent forces may contribute to apical movement during division, but the length of the cell may determine which mechanism contributes the most. Highly elongated cells, such as radial glia, may require the microtubule cytoskeleton. Shorter epithelia undergo shorter nuclear movements to and from the apical surface. In this case, actin-based forces that round up the cell may suffice, with little need for the microtubule cytoskeleton. Cytoplasm of the cells is shown in green, nuclei and DNA are shown in blue. Centrosomes are red and cilia are magenta.
Download figure to PowerPoint
Initiation of mitosis in the neuroepithelium
How mitosis initiates in the neuroepithelium is unclear. Apically positioned centrosomes are presumed to trigger mitosis only after nuclei have reached the apical surface (Hinds & Ruffett 1971; Smart 1972; Frade 2002; Miyata 2007; Tamai et al. 2007;Schenk et al. 2009; Taverna & Huttner 2010). The centrosome is critical for entry into mitosis. Mitosis is triggered by a cascade of proteins, culminating in the activation of aurora A kinase, which in turn activates the cyclin B/cdk1 complex. This cascade is localized to the centrosome (Hirota et al. 2003; Jackman et al. 2003). Centrosomes are important for nuclear envelope breakdown (NEBD) as well (Margalit et al. 2005; Basto & Pines 2007). However, it has long been noted that early prophase cells were found away from the apical surface, suggesting that mitosis may not initiate at the apical surface (Sauer 1936; Hinds & Ruffett 1971). Reduction of Lis1, loss of microtubule stability, and inhibition of actin produce ectopic mitotic cells in the neuroepithelium (Murciano et al. 2002; Gambello et al. 2003; Xie et al. 2007), suggesting that either mitosis does not initiate at the centrosomes or that centrosomes do not stay at the apical surface. We have observed that the latter is true. We have previously reported that in the dorsal neural tube, centrosomes leave the apical surface prior to mitosis (Ahlstrom & Erickson 2009). More recently, we have discovered that late in G2 phase, centrosomes of the lateral neural tube and cerebral cortex leave the apical surface and contact the nucleus to initiate mitosis, which we discuss in the next section.
Centrosome positioning and initiation of mitosis
Centrosome positioning is critical for normal INM since disruption of centrosome position perturbs INM. For example, loss-of-function mutations of atypical protein kinase C λ (aPKCλ) cause disorganization of cell polarity and displace centrosomes from the apical surface in the neuroepithelium. Subsequently, mitosis occurs non-apically (Imai et al. 2006). It has also been observed that loss of Pax6 is associated with abnormal INM. In this circumstance, the centrosomes move away from the apical surface early in the cell cycle, and are frequently found in ectopic locations (Tamai et al. 2007).
How centrosomes are localized to the apical surface is not certain. Centrosomes are physically connected to the cilium. Therefore, it is likely that the centrosome is anchored at the apical surface via a primary cilium (Dubreuil et al. 2007). Consistent with this idea, when Pax6 is reduced in the neuroepithelium, cells often lack a primary cilium and under these circumstances centrosomes move away from the apical surface. We have observed in the chicken neural tube that cilia are disassembled late in G2 phase. Concomitant with this event, centrosomes depart from the apical surface before the incoming nucleus reaches the apical surface (Fig. 2). When the centrosomes arrive at the nucleus, nuclear envelope breakdown proceeds. In 66% of mitoses observed in the chicken neural tube, nuclear envelope breakdown began more than 7 μm from the apical surface.
Figure 2. Centrosomes leave the apical surface to initiate mitosis. The intact nucleus is marked with NLS-tdTomato and an asterisk; centrosomes are marked with GFP-centrin and green arrows. The average distance from site of nuclear envelope breakdown to final position during mitosis is 16.5 μm (n = 15). All mitotic cells are observed to undergo apical rounding during mitosis (n = 62). NLS-tdTomato and GFP-centrin were introduced to neural tube cells by in-ovo electroporation at Hamburger/Hamilton stage 12–16. Neural tubes were sectioned and imaged <24 h later on an Olympus confocal microscope. Images were taken every 7 min. Scale bar = 10 μm.
Download figure to PowerPoint
As with apical nuclear movement, there is also disagreement about the mechanism of basally-directed nuclear movement during INM. Schenk et al. (2009) suggest that myosin II and actin control basal nuclear movement in the mouse cortex, whereas Tsai et al. (2010) propose that nuclei are driven along microtubules via kinesin 3. Yet a third possibility initially suggested by FC Sauer, but more recently supported experimentally by Kosodo et al. (2011), is that basal nuclear movement during INM is not an active, cell-autonomous process (Sauer 1935). When magnetic beads are implanted into the neuroepithelium, they are observed to move basally. However, if apical nuclear migration is blocked by inhibiting S-phase with hydroxyurea, beads fail to move basally, suggesting that basal displacement is passively driven by active apical nuclear movement (Kosodo et al. 2011).
As with apical nuclear movement, these mechanisms need not be mutually exclusive. Myosin II was not shown to be required cell autonomously for basal nuclear movement because tissue sections were incubated in blebbistatin, a myosin inhibitor, which affected all cells (Schenk et al. 2009). Cells in mitosis moving to the apical surface may require myosin II to round up forcefully and consequently displace their neighbors basally. Based on the fact that kinesin 3 and myosin both seem essential for basal nuclear migration, we suggest that kinesin 3 may be required in G1 cells as a ratchet that prevents back-sliding and therefore allows effective movement in one direction only, rather than as an active motor. This would ensure that pressures from rounding neighboring mitotic cells would move nuclei toward the plus ends of microtubules, and away from the apical surface, instead of pushing the nucleus basally and apically randomly. Alternatively, intrinsic microtubule-based nuclear movement in G1 driven by kinesin may be required for persistent basal nuclear movement after neighboring cells have already pushed G1 nuclei away from the apical surface.
Relationship of cell cycle and nuclear movement
Cell cycle is correlated with nuclear movement. Mitotic nuclei are found at the apical surface for reasons previously discussed. S-phase cells are found at more basal positions. Time-lapse imaging combined with S-phase labeling suggest that G1 nuclei are migrating basally, whereas G2 nuclei are migrating apically (Hayes & Nowakowski 2000; Tamai et al. 2007). Cell cycle progression appears to be essential for nuclear migration. Inhibiting completion of mitosis with 5-azacytidine prevents basal nuclear movement, as evidenced by an increase of mitotic figures at the apical surface (Ueno et al. 2006).
There is more controversy about whether the completion of S-phase is necessary for apical nuclear migration. In support of this, overexpression of p18Ink4c arrests cells in G1 phase, and results in an increase of cells in the basal region of the neuroepithelium (Kosodo et al. 2011). Also, morphine is known to slow down the cell cycle of radial glial cells and delays apical nuclear migration (Sargeant et al. 2008). Finally, Ueno et al. (2006) found that cyclophosphamide-induced arrest in S-phase prevents apical nuclear migration, and Leung et al. (2011) demonstrated that arresting retinal cells with aphidicolin and hydroxyurea inhibits apical nuclear migration (Kosodo et al. 2011). Curiously, Murciano et al. (2002) reported that inhibition of S-phase with hydroxyurea does not inhibit apical nuclear migration in the chicken diencephalon. The discrepancy between their results and others detailed above is not explained.
How cell cycle might regulate nuclear movement is not known. A possible link between the S-to-G2 phase transition and apical nuclear movement is TPX2, which is contained in the nucleus prior to G2 phase, and relocates to the apical process during G2 phase. TPX2 is a microtubule-binding protein that reorganizes the microtubule cytoskeleton, and is essential for apical nuclear movement (Kosodo et al. 2011). Potentially, INM is controlled by a cell-cycle-regulated release of TPX2.