Cell biology of pseudostratified epithelia

Cells within a PSE share many features with other epithelial cells. They are attached to the apical and basal lamina via apical and basal processes (see poster). Furthermore, they show clear apico-basal polarity (Drubin and Nelson, 1996), and apical polarity markers, including the Par complex, ZO-1 (also known as TJP1) and Crumbs proteins are found at their apical membrane. PSE feature apical junctions, namely septate junctions in Drosophila and adherens and tight junctions in vertebrates (Liang et al., 2014; Strzyz et al., 2015). In all PSE, the centrosome is found at apical positions during the entire cell cycle (see poster). While this is most likely linked to the fact that the centrosome serves as a basal body for the apically emerging primary cilium in vertebrates (Kosodo et al., 2011; Leung et al., 2011; Meyer et al., 2011; Norden et al., 2009; Okamoto et al., 2014; Tsai et al., 2010), the relevance of the apical centrosome position in Drosophila PSE that lack primary cilia (Meyer et al., 2011) is less clear. In interphase, microtubules emanate from this apical centrosome meaning that all microtubule plus-ends point basally (Kosodo et al., 2011; Leung et al., 2011; Meyer et al., 2011; Norden et al., 2009; Okamoto et al., 2014; Tsai et al., 2010). Only upon entry into mitosis, the duplicated centrosomes leave apical positions and travel basally to meet the nucleus. They then guide the formation of a bipolar spindle that is arranged perpendicular to the apical surface (Dzafic et al., 2015; Strzyz et al., 2015).

Like the centrosome, the Golgi complex is also confined to the apical process in cells of zebrafish neuroepithelia, as well as the mouse neocortex PSE; however, Golgi and centrosome are not connected to each other (Hinds and Hinds, 1974; Taverna et al., 2016). In contrast, the endoplasmic reticulum can be found in both apical and basal processes (Taverna et al., 2016). The same appears to be true for mitochondria (my unpublished observations) and many types of endosomes (Clark et al., 2011). While the cell biology of the apical process is well defined (see poster), much less is known about the composition of the basal process, which is linked to the basal lamina through integrin-based focal adhesion complexes (Stuermer and Bastmeyer, 2000). Interestingly, in the zebrafish retinal neuroepithelium, it appears that these basal adhesions are more stable than the adhesions of emerging neighbouring neurons (Icha et al., 2016). It has further been shown that regrowth of the basal process is more directed and faster than regrowth of the apical process (Icha et al., 2016). However, the reasons for these differences in adhesion or growth potential are not yet clear.

Pseudostratified epithelia heterogeneity

As noted above, cells in PSE share many cell biological characteristics independently of where they occur; however, PSE are a heterogeneous group of tissues. While all PSE show layered nuclear arrangements with nuclei dispersed along the apico-basal axis, they vary extensively in height, spanning distances of less than 30 µm in the Drosophila optic lobe (Rujano et al., 2013) to hundreds of micrometres in the rodent neocortex (Tsai et al., 2010), even reaching millimetres in the primate neocortex (Rakic, 1972). Alongside an increase in epithelial height, the number of nuclear layers stacked within a PSE also increases and, concurrently, so does the distance the nuclei have to travel to reach apical positions before mitosis (also see the images on the poster). Depending on the cell length and number of nuclear layers, I will here categorize PSE into short, intermediate and extended PSE (see poster) (Strzyz et al., 2016).

Short PSE (below 30 µm) are seen in both invertebrates and vertebrates. For example, cells in the optic lobe in Drosophila that give rise to parts of the fly eye are less than 30 µm high, and nuclei are arranged basically in a top and a bottom layer (Rujano et al., 2013). Similar PSE heights are found in some vertebrate tissues, including endodermal organ buds (Bort et al., 2006), otic vesicles (Hoijman et al., 2015), the neural plate (Schoenwolf and Alvarez, 1989) and zebrafish photoreceptor cell precursors (Weber et al., 2014).

Intermediate PSE span an apico-basal distance of between 30 µm and 60 µm, with nuclei stacked into four to six layers. Consequently, cells are less columnar and become more spindle shaped (see poster). In turn, both, apical and basal processes are thinner than in short PSE with less cytoplasm and a denser packing of nuclei. Examples of such intermediate epithelia are fly imaginal discs (Meyer et al., 2011), Nematostella ectoderm (Meyer et al., 2011), zebrafish retinal and hindbrain neuroepithelia (Leung et al., 2011; Norden et al., 2009), mouse intestine (Grosse et al., 2011) and the mammalian neural tube (Sauer, 1935).

Extended PSE that span ∼100 µm and feature eight to nine nuclear layers do exist, but they are less common. Examples are the more developed neural tube and retinal neuroepithelia in mammals (Iulianella et al., 2008). One particular case of an extended PSE is the PSE of mammalian neocortices that consist of radial glia cells (see poster). These cells are extremely elongated, spanning about 250 µm in mouse and up to 1 mm in primates (Miyata, 2008; Rakic, 1972). In these cells, the only bulky region is the nucleus, while processes are extremely thin and almost devoid of any cytoplasm. This gives radial glia cells a ‘bead on a string’ appearance. In contrast to other PSEs, here nuclei are not found along the entire apico-basal axis, but instead reside in a restricted zone that spans ∼150 µm from the apical surface where they arrange into ten or more layers.

In addition to their differences in length and nuclear layer arrangements, PSE can also vary in their overall shape and architecture. The majority of PSE have a straight surface and a rectangular morphology, as seen in the ectoderm of Nematostella and PSE in the neocortex. However, some PSEs are apically or basally bent, for example parts of the Drosophila wing disc and the vertebrate retina (see poster). Whether this different architecture is linked to differences in nuclear arrangements and packing states, or even influences how nuclei move within epithelia has not yet been explored. Furthermore, it is not fully understood whether differences in tissue height or tissue packing can influence cellular processes within PSE (see poster). While many of these areas still need investigations, some correlations are found with regard to the observed links between epithelial length and the machineries that drive nuclear dynamics during IKNM (Lee and Norden, 2013; Miyata et al., 2015; Strzyz et al., 2016), which will be discussed and elaborated in the next chapter.

Machineries driving nuclear movements in PSE – possible dependence on tissue context

IKNM is the most studied hallmark of PSE. For almost a century, researchers have investigated why and how nuclei move in PSE during the cell cycle. The phenomenon of IKNM was already discovered in the 1930s when Sauer noted that all cell divisions occur at apical positions and nuclei vacate this space in other phases of the cell cycle (Sauer, 1935). I will not elaborate on the basal movements of nuclei here, as they have been discussed in detail in previous reviews (Lee and Norden, 2013; Miyata et al., 2015; Strzyz et al., 2016), but focus on the apical nuclear movement before mitosis.

It is now clear that the apical nuclear movement that brings nuclei to the apical surface before mitosis is a fast, directed and active process that is driven by cytoskeletal elements in all PSE. However, the molecular machineries responsible for these movements can vary depending on tissue context. So far, a combination of studies hints that the length of a PSE might be linked to the particular molecular machinery used.

In PSE of short and intermediate length, actin and myosin have been shown to be involved in apical IKNM (Leung et al., 2011; Meyer et al., 2011; Norden et al., 2009; Rujano et al., 2013). Nevertheless, differences are seen in how actin and myosin are used. In the short PSE of the Drosophila optic lobe, for example, where nuclei have to travel only one nuclear length to reach the apical surface, it appears that actomyosin-based cell rounding is sufficient to move nuclei into apical positions for mitosis. In line with this argument, in these epithelia, myosin is seen only at the basal cortex, and contractility of this actomyosin assembly, which appears to occur concurrently with cell rounding, might be sufficient to move nuclei to apical positions (see poster) (Rujano et al., 2013). A similar movement is seen in zebrafish photoreceptor cell precursors, which have a comparable length to cells of the optic lobe (Weber et al., 2014). However, here, the cytoskeletal arrangements involved have not yet been explored.

In PSE of intermediate length, such as the zebrafish retinal neuroepithelium or the imaginal disk of the fly wing, mitotic rounding and IKNM can occur independently of each other, indicating that, here, IKNM is an additional process. Interestingly, mitotic rounding can take place at basal positions in the epithelium in cases where IKNM is impaired due to genetic manipulation or interference with the actin pool that drives the movement (Liang et al., 2014; Strzyz et al., 2015). This argues that in these epithelia, in which nuclei have to travel up to three times their length to reach the apical surface, additional mechanisms other than cell rounding that are based on actin and myosin have evolved. The finding that myosin and actin follow the nucleus as it moves apically underlines this notion (Meyer et al., 2011; Norden et al., 2009). However, how exactly actin and myosin generate force to translocate the bulky nucleus in these tissues is still not well understood. In the fly wing, Rho-associated protein kinase (Rock)-dependent mechanisms have been suggested, implying that actomyosin contractility could be the driving force (Meyer et al., 2011). For a brief discussion of the role of microtubules in these tissues, see Box 1.

Finally, in the very elongated radial glia cells of the neocortex, in which nuclei have to move up to ten times, or even more, their own length to reach apical positions, microtubules and their associated motors are actively involved in apical nuclear movement before mitosis (see poster). Microtubules are arranged as apico-basal arrays that form a cage surrounding the nucleus (see Box 1) (Xie et al., 2007). If this cage is compromised by either depolymerising microtubules or by interference with centrosome integrity, apical movements are disturbed (Ge et al., 2010; Xie et al., 2007). It has been shown that dynein serves as an apically-directed motor and the plus-end-directed kinesin motor Kif1A moves nuclei basally in the rodent neocortex (Tsai et al., 2010). Dynein is recruited to the nuclear envelope by the nuclear pore complex proteins Nup133 and BicD2, which then link the nucleus to the microtubule network through dynein (Hu et al., 2013).

Overall, the data acquired to date could suggest that the observed switch from an actomyosin-based force-generating mechanism to one that involves microtubule-mediated transport is connected to tissue height. If this speculation is correct, actin and myosin might not be able to generate sufficient force to move nuclei to apical positions as overall length increases, and microtubule-based transport could be a more robust mechanism for extremely elongated cells. So far, however, this idea is purely speculative. Future studies using newly established 3D culture models for PSE, including organoids, could help to investigate this question in more detail.

Apical IKNM and its significance for tissue integrity

In addition to elucidating how nuclei move apically before mitosis it is also important to understand why mitosis and cell division always occur at apical positions in the PSE (see poster). One hypothesis, initially raised in the 1970s, is that rounded mitotic cells occupy more space and moving them to the apical surface could thus ‘outsource’ them from the rest of the epithelium (Fish et al., 2008; Smart, 1972). However, it is only in outward curved PSE, such as the retinal neuroepithelium, that the apical side indeed has a greater apical than basal surface area (see poster). Therefore, in the majority of straight PSE, having a larger surface area available for mitotic events is most likely not the main reason for apical mitosis. In addition, non-apical mitosis can occur despite dense nuclear packing when apical IKNM is disturbed or when committed precursors arise in later retinal development (Liang et al., 2014; Strzyz et al., 2015; Weber et al., 2014); this further argues that eventual space constraints are not the main reasons for apical mitosis.

Another hypothesis has suggested an apical ‘mitotic zone’ is present in the very elongated rodent neocortex PSE. Here, it was proposed that apical signalling cues allow for a spatial control of mitotic entry (Hu et al., 2013). So far, the exact nature of these apical signals remains elusive and needs further investigation. It is possible that such mitotic entry signals are specific to extended PSE, as, as mentioned above, in shorter PSE, cells can enter mitosis at non-apical positions (Liang et al., 2014; Strzyz et al., 2015; Weber et al., 2014). Another consideration is that the apically positioned centrosome could be instructive for apical IKNM. Apical nuclear movement before mitosis could facilitate the interaction between the centrosome and nucleus, which is important for bipolar spindle formation. This might well be true in the very elongated cells of the neocortex PSE, in which the centrosome is also actively involved in nuclear movement. However, apical IKNM in the zebrafish retinal PSE takes place even when the centrosome is mispositioned or ablated (Strzyz et al., 2015). In addition, in PSE of the Drosophila wing disc, apical IKNM is still observed when centrosomes are genetically depleted (Poulton et al., 2014). Thus, centrosome position, and even its presence, can be decoupled from apical IKNM, at least in some tissues.

In most PSE investigated so far, apical IKNM is followed by a cell division that features a perpendicular cleavage plane (see poster) (Cui et al., 2007; Das et al., 2003; Kosodo et al., 2004; Nakajima et al., 2013). As such divisions bisect the apical membranes, correct cleavage angles could be important for the distribution of apical components into the two daughter cells. In some PSE, including the Drosophila wing disc, the chick neural tube and the mouse neocortex, a perturbation of cleavage plane orientation can result in cell delamination and a misregulation of cell fates (Morin et al., 2007; Nakajima et al., 2013). In the retinal neuroepithelium, however, a perpendicular cleavage plane is not absolutely necessary for cells to re-integrate into the tissue (Dzafic et al., 2015; Strzyz et al., 2015). Here, as long as divisions still occur in apical regions, both daughter cells will be included in the tissue and continue to proliferate. However, when apical IKNM is perturbed in this tissue (Strzyz et al., 2015), and cells divide at more basal locations, at least one of the daughter cells does not re-establish bipolar PSE morphology. Such a cell ectopically proliferates and interferes with tissue integrity and consecutive retinogenesis (Strzyz et al., 2015) (see related images on the poster). Taken together, these studies show that an important role of apical IKNM is to ensure apical divisions, which in turn are necessary and important for daughter cells to re-integrate into the tissue, and thus ensure tissue integrity before differentiation and maturation.