Synchronisation of apical constriction and cell cycle progression is a conserved behaviour of pseudostratified neuroepithelia informed by their tissue geometry

Neuroepithelial cells balance tissue growth requirement with the morphogenetic imperative of closing the neural tube. They apically constrict to generate mechanical forces which elevate the neural folds, but are thought to apically dilate during mitosis. However, we previously reported that mitotic neuroepithelial cells in the mouse posterior neuropore have smaller apical surfaces than non-mitotic cells. Here, we document progressive apical enrichment of non-muscle myosin-II in mitotic, but not non-mitotic, neuroepithelial cells with smaller apical areas. Live-imaging of the chick posterior neuropore confirms apical constriction synchronised with mitosis, reaching maximal constriction by anaphase, before division and re-dilation. Mitotic apical constriction amplitude is significantly greater than interphase constrictions. To investigate conservation in humans, we characterised early stages of iPSC differentiation through dual SMAD-inhibition to robustly produce pseudostratified neuroepithelia with apically enriched actomyosin. These cultured neuroepithelial cells achieve an equivalent apical area to those in mouse embryos. iPSC-derived neuroepithelial cells have large apical areas in G2 which constrict in M phase and retain this constriction in G1/S. Given that this differentiation method produces anterior neural identities, we studied the anterior neuroepithelium of the elevating mouse mid-brain neural tube. Instead of constricting, mid-brain mitotic neuroepithelial cells have larger apical areas than interphase cells. Tissue geometry differs between the apically convex early midbrain and flat posterior neuropore. Culturing human neuroepithelia on equivalently convex surfaces prevents mitotic apical constriction. Thus, neuroepithelial cells undergo high-amplitude apical constriction synchronised with cell cycle progression but the timing of their constriction if influenced by tissue geometry.


Introduction
Embryonic cells must balance physiological requirements for growth with energy-intensive force generating behaviours necessary to form functional organs during morphogenesis. Closure of the neural tube is a clinically relevant paradigm of morphogenesis. It requires bending of the neural plate, which elevates paired neural folds that meet at the dorsal midline, forming closure points from where "zippering" initiates to progressively close the open neuropores . Failure to close the neural tube causes neural tube defects in an average of 1:1000 births globally (Zaganjor et al., 2016). Incomplete closure of the anterior neuropores causes fatal anencephaly whereas failure to close the posterior neuropore (PNP) causes spina bifida. Both cranial and spinal neural tube closure require neuroepithelial cells to undergo apical constriction (Butler et al., 2019;Galea et al., 2021;Lesko et al., 2021;Kowalczyk et al., 2021).
Apical constriction is a conserved force-generating behaviour whereby actomyosin recruitment shrinks the apical domain of epithelial cells. It has been intensively studied in Drosophila, Xenopus and nematodes, identifying common and distinct constriction mechanisms (Martin and Goldstein, 2014). Some cells preferentially recruit actomyosin to the apicomedial cell surface, assembling radial F-actin filaments which guide constriction towards the middle of the cell's apex (Coravos and Martin, 2016). Other cell types preferentially recruit myosin to the cell cortex, serially and often directionally shrinking cell junctions (Nishimura et al., 2012). Both apicomedial and cell junction actomyosin can contribute to apical shrinkage and much of the contractile machinery is common to both constriction modalities. Both require non-muscle myosin motor proteins such as myosin-IIb, activated through phosphorylation by kinases including Rho-associated kinase (ROCK). Phosphorylated myosins localise to the cell apical surface where they bind F-actin scaffolds and generate contractile forces through ATP-dependent power strokes. In the stratified neuroepithelium of Xenopus, apicomedial F-actin accumulation is associated with apical shrinkage in the presumptive brain, whereas both apicomedial and cortical accumulation can correlate with apical shrinkage in the spinal region (Baldwin et al., 2022).
We have documented both apicomedial and cortical myosin-IIb localisation in the pseudostratified neuroepithelium of the mouse PNP during spinal neurulation . Apicomedial myosin-IIb localisation correlates with dense microtubule networks in both interphase and mitotic cells, in which they are apical to, and distinct from, the mitotic spindle microtubules . Cortical localisation of myosin heavy chain (MHC)-IIb is diminished in non-constricting neuroepithelial cells whose neighbours lack the planar cell polarity (PCP) co-receptor VANGL2. ROCK is a downstream effector of PCP signalling known to enhance neuroepithelial apical constriction in multiple species, including the mouse and chick (Butler et al., 2019;Nishimura et al., 2012). Its pharmacological inhibition globally reduces neuroepithelial tension, enlarges apical surfaces and diminishes apical phospho-myosin-IIb localisation in the mouse PNP (Butler et al., 2019;Escuin et al., 2015). We previously reported that ROCK's effects on apical constriction are related to cell cycle progression. As neuroepithelial cells proliferate in the presence of ROCK inhibition, the proportion of cells with dilated apical surfaces increases and the PNP progressively widens (Butler et al., 2019). This demonstrates the importance of coordination between apical constriction and cell proliferation. Excessive neuroepithelial proliferation has repeatedly been associated with failure of neural tube closure in mouse genetic models (Lardelli et al., 1996;Anderson et al., 2016;Parchem et al., 2015;Badouel et al., 2015).
Cell cycle progression in the highly proliferative neuroepithelium is unusual because these cells are pseudostratified in amniotes. Consequently, their nucleus displaces along the apical-basal axis due to interkinetic nuclear migration (IKNM) as cells progress through the cell cycle. Mitoses are predictably localised apically, followed by stochastic basal descent (Leung et al., 2011). As cells pass through G2 phase their nuclei rapidly ascend apically. G2 nuclear ascent is an active process which can be achieved by different cellular mechanisms in tissues with different geometries. For example, in the flat neuroepithelium of the zebrafish hindbrain it requires ROCK-dependent myosin activation, whereas in the apically convex developing retina it requires formin-mediated sub-nuclear F-actin accumulation (Yanakieva et al., 2019a). These displacements require nuclear deformation, which is opposed by nuclear envelope proteins such as Lamin A/C (Swift et al., 2013). Embryonic neuroepithelia characteristically lack A-type Lamin expression (Jung et al., 2012;Takamori et al., 2018;Yanakieva et al., 2019a) and over-expression of Lamin A diminishes their nuclear motility during IKNM (Yanakieva et al., 2019b).
IKNM is believed to reverse neuroepithelial apical constriction by causing apical dilation during mitosis (Guerrero et al., 2019). The dogma that mitosis causes apical expansion is strongly supported by findings in Drosophila showing reversal of apical constriction and loss of adherens junctions during mitosis (Ko et al., 2020;Aguilar-Aragon et al., 2020). However, it does not hold true in the mouse PNP neuroepithelium: we previously reported that neuroepithelial cells labelled with proliferating cell markers such as phospho-histone H3 (pHH3) unexpectedly have smaller apical areas than interphase cells (Butler et al., 2019). We therefore propose that synchronisation of apical constriction with cell cycle progression is a neuroepithelial behaviour which reconciles physiological growth requirements with the necessity of closing the neural tube.
As a recently described cell behaviour, very little is known about the regulation of neuroepithelial mitotic apical constriction. Here, we set out to document its relation to apical myosin recruitment, timing during mitotic progression, and conservation between species. We document that synchronisation of apical constriction with cell cycle progression is a conserved property, including in human cells differentiated from induced pluripotent stem cells (iPSCs). However, whereas flat neuroepithelia undergo mitotic apical constriction, mitotic dilation predominates in apically convex neuroepithelia of the early mouse midbrain and human cells cultured on convex surfaces.

Mouse posterior neuropore neuroepithelial cells accumulate apical myosin IIb
Neuroepithelial cells have tortuous shapes with numerous sub-apical protrusions (Kasioulis et al., 2022), variable nuclear positioning along the apicobasal axis and often unpredictable regional widening or narrowing of the cell body (Fig. 1A). Here, we quantify apical surface properties related to each cell's body or nucleus using 3D visualisation aided with basolateral markers (scribble or adherens junction markers). Individual cells can be traced to relate deep nuclei to the cell's apical surface as previously described (Butler et al., 2019). Apically located mitotic chromatin does not protrude beyond the PNP epithelial surface, rather sitting below a constricted cell apex (Fig. 1A) frequently decorated with ROCK (Fig. 1B).
MHC-IIb is localised in apicomedial pools ("cap") and/or circumferentially at cell-cell junctions ("cortical",Fig. 1C and D). Mitotically rounded cells continue to localise MHC-IIb around their apical cap and/ or cortex similarly to non-mitotic cells, as well as in their sub-apical cell cortex ( Supplementary Fig. 1). Individual cells' apicomedial cap and cortical MHC-IIb intensities are correlated in both mitotic and nonmitotic cells (mitotic R 2 ¼ 0.51, p < 0.001; non-mitotic R 2 ¼ 0.48, p < 0.001). Neither apical cap nor cortical MHC-IIb intensity is correlated with the apical area of interphase (pHH3-) cells (Fig. 1E). In contrast, both myosin pools are inversely correlated with apical area in mitotic cells (pHH3þ, Fig. 1E), suggesting that progressive accumulation of apical myosin predictably shrinks their apical surface. This association may not be evident in static analyses of interphase cells due to the pulsatile nature of apical constriction (Christodoulou and Skourides, 2015;Galea et al., 2021). Cells with equivalent apical areas may have been either constricting or dilating at the point of embryo fixation (Fig. 1F).

Mitotic apical constriction is the highest magnitude constriction in the chick spinal neuroepithelium
The chick spinal neuroepithelium is more amenable to dynamic analyses requiring long-term live imaging than the mouse. Species differences are notable: average apical areas are smaller in the chick than mouse PNP neuroepithelium ( Fig. 2A and B). Nonetheless, mitotic neuroepithelial cells have smaller apical areas than their corresponding interphase cells in both chick and mouse embryos (Fig. 2B), confirming evolutionary conservation of mitotic apical constriction. Apical areas of mouse mitotic neuroepithelial cells were on average 40.7% smaller than non-mitotic cells (pHH3 þ 20.5 μm 2 versus pHH3 -34.5 μm 2 ) and those of chick embryos decreased by 33% (pHH3 þ 13.5 μm 2 versus pHH3 -19.2 μm 2 ).
The chick neuroepithelium can be live-imaged with vital cell dyes through a window in the vitelline membrane, which does not impair embryo growth, proliferation or mitotic apical constriction over experimentally relevant timeframes . DiI labelling, optimised to produce sparse mosaic labelling (Supplementary Figs. 2E-F), enables individual cells to be tracked over time (Supplementary Fig. 2G). Individual cells asynchronously displace their nuclei and either constrict or dilate during imaging (Fig. 2C). We used nuclear morphology to identify cells which progressed through mitosis during live imaging, versus those that remained in interphase throughout (Fig. 2D) and CellMask membrane stain to relate each nucleus or mitotic chromatin to its apical surface (Fig. 2E).
Live-imaged chick neuroepithelial cells undergo apical constriction in the 30 min preceding anaphase (Fig. 2F). Both mitotic and interphase apical constrictions follow a pulsatile pattern (Fig. 2G). To reconcile asynchronous changes, individual cell profiles were temporally aligned. T0 was defined as the timepoint when anaphase morphology was observed for mitotic cells and the smallest recorded apical area for nonmitotic cells. Individual cell traces were normalised to each cell's average apical area to reveal trends independently of inter-cell variability ( Fig. 2G and H). Mitotic cells tend to dilate up to approximately 30 min prior to anaphase, then predictably constrict to a minimum apical area at anaphase before gradually re-dilating to an interphase-like pattern within 30 min following anaphase (analysing one daughter cell following cytokinesis) (Fig. 2G). Following division, the daughter cells resume a pattern of lower-magnitude apical area fluctuations equivalent to those seen in cells which do not divide during live imaging. The pulsatile nature of individual cells' apical constrictions can be approximated by plotting a sine curve, from which constriction pulse amplitude can be calculated (Fig. 2H). The constriction amplitude of mitotic cells is significantly greater than apical area oscillations observed in interphase (Fig. 2I).

Human iPSC-derived neuroepithelia apically constrict between G2 and M phase
Equivalent analyses are not possible in human embryos. Many protocols have been described to differentiate human iPSCs into neural progenitor cells, but few have been extensively characterised in the early stages of neuroepithelial induction to confirm transition through this morphogenetically-relevant stage. Additionally, most protocols induce a presumptive neuroepithelial stage as a cell aggregate with deep apical surfaces inaccessible to en face imaging. We sought to characterise a human system producing flat neuroepithelial sheets with accessible apical surfaces. Dual-SMAD inhibition (Shi et al., 2012b;Chambers et al., 2009) is well known to produce neurogenic cells in a flat sheet (Fig. 3A), but its early differentiation stages are minimally characterised.
Over 8-days of differentiation these cells lose pluripotency markers including NANOG (Fig. 3A), but retain the dual iPSC and neuroepithelial marker SOX2 while gaining the neuroepithelium-selective adherens junction marker CDH2 ( Fig. 3B and C). A molecular peculiarity of neuroepithelial cells is their loss of Lamin A/C while retaining Lamin B around their nuclear envelope, both of which are recapitulated in these cultures ( Fig. 3B and C). This minimum panel of molecular markers defines a NANOG -/SOX2 þ /CHD2 þ /Lamin A/C À epithelial population which, to our knowledge, is exclusive to neuroepithelial cells, complementing previously reported (Chambers et al., 2009;Shi et al., 2012b) extensive molecular characterisation of dual-SMAD differentiated cells. Of greater consequence to the current studies is the morphology of the epithelial layer produced. Following eight days of differentiation, iPSCderived neuroepithelial cells are apicobasally polarised with apical localisation of ZO-1 and enrichment of phospho-myosin, adhered onto a single continuous basement membrane containing laminin (Fig. 3D). Cell nuclei are dispersed between these discrete apical and basal domains, forming approximately three pseudo-rows ("nuclear index", Supplementary Figs. 3A-B). Each cell spans the apical to basal domain of the epithelium, which is as thick as the E9.5 mouse neuroepithelium ( Fig. 3F and G). They form sub-apical lateral protrusions (Fig. 3E) similar to those recently reported in vivo (Kasioulis et al., 2022). Thus, this short protocol of human iPSC differentiation produces a tractable neuroepithelial sheet through which to mechanistically study cell behaviours such as apical constriction.
Both iPSC lines tested achieve apical areas equivalent to neuroepithelial cells in the mouse PNP (Fig. 4A). Mitotic cells localise their rounded body apically, under a small apical surface (Fig. 4B). In these human cells, Ki-67 can be used to segregate cells in different stages of the cell cycle as previously reported (Galea et al., 2013;Ghule et al., 2011). Ki-67 forms multiple small nuclear foci in early G1 which expand to form two condensates in S, coalescing into a single nuclear punctum in G2 and then spreading over the condensing chromosomes as cells enter M phase ( Fig. 4C and Supplementary Fig. 3C). We confirmed that the Ki-67 staining pattern correlates with nuclear volume, as a surrogate of cell cycle progression (Chu et al., 2017), in iPSC-derived neuroepithelial cells. Cells in early G1 have the smallest nuclear volume, increasing during genome duplication to a maximum average volume in G2 (Supplementary Figs. 3C-E).
In vivo, G1 and S phase nuclei can be distributed throughout the neuroepithelium, whereas mitotic chromatin is predictably apical (Baye and Link, 2007). Human iPSC-derived mitotic neuroepithelial cells apically localise their chromatin below a ring of phosphorylated myosin ( Fig. 4D). Both human iPSC lines tested show variations of apical area of individual cells in each stage of the cell cycle, demonstrating that apical area variability is an inherent feature of neuroepithelia even in the absence of extrinsic chemical or mechanical inputs (Fig. 4E). Both lines also show significantly larger apical areas in the G2 phase of the cell cycle, constricting to significantly smaller areas in M phase and retaining equivalent areas in G1/S (Fig. 4E). Thus, G2 to M phase apical constriction is an intrinsic neuroepithelial behaviour conserved in human cells.

Tissue geometry influences the timing of apical constriction during cell cycle progression
The demonstration of apical constriction in different species prompted us to test conservation of this behaviour in a different anatomical site, namely the anterior neuroepithelium of mouse embryos (Fig. 5A). Contrary to the spinal region, anterior neuroepithelial cells in the presumptive midbrain do not undergo mitotic apical constriction, but rather dilate their apical surfaces, producing the largest apical areas observed in the epithelium (Fig. 5B-D). MHC-IIb apical localisation shows both cortical and apical cap localisation in the midbrain region analysed (Fig. 5E), as seen in the PNP. However, contrary to the PNP, no correlation is present between apical area and myosin staining intensity in either pHH3þ (Fig. 5F) or pHH3-( Supplementary Fig. 3A) anterior neuroepithelial cells. Consistent with this, ROCK is not notably enriched around the apical cortex of mitotic anterior neuroepithelial cells ( Supplementary  Fig. 3B). Given that pHH3 labels cells from G2 to anaphase, we assessed whether the proportion of cells with corresponding nuclear morphologies differ between the anterior and spinal neuroepithelium, finding no differences ( Supplementary Fig. 4A). However, we observed that pHH3þ nuclei with a G2 morphology were on average closer to the neuroepithelial apical surface in the anterior than spinal neuroepithelium (Supplementary Figs. 4B and C).
Nuclear position in fixed embryos is a snapshot of their dynamic displacement apically. The molecular mechanisms and rate controlling this ascent have previously been shown to be influenced by neuroepithelial curvature (Yanakieva et al., 2019a;Ishii et al., 2021), which is substantially different between the relatively flat PNP versus the apically convex midbrain region analysed (Fig. 5G). We therefore hypothesised that tissue geometry may determine whether neuroepithelial cells undergo mitotic apical constriction or dilation. Consistent with this, the median apical area of midbrain mitotic cells in individual embryos is significantly negatively correlated with their tissue's radius of curvature: greater tissue curvature correlates with larger apical areas of mitotic cells within the same embryo (Supplementary Figs. 4D and E).
To directly test the effect of geometry on mitotic apical constriction, we differentiated human iPSC-derived neuroepithelia on coated spherical glass beads mimicking early midbrain curvature ( Fig. 6A and B). After seeding, cells formed a dense lawn over the beads, expanding to form a thick layer along the bead sides and joining adjacent beads distinctly visible within five days of seeding (Fig. 6A). Cells on the top of these beads did not form a pseudostratified epithelium and were excluded from further analyses, but those along the sides of the beads form a thick, apically convex neuroepithelium ( Fig. 6C and D, Supplementary Fig. 6). Their resulting apical curvature is variable, but on average is at least as curved as that observed in the early mouse midbrain (Fig. 6E). Cells in these apically convex regions continue to apically localise cortical phospho-myosin (Fig. 6F). As had been seen in flat Quantification of neuroepithelial thickness on day 2 and day 8 of neuroepithelial differentiation. Points represent independent experiments, linked by black lines. P value by paired t-test. The dashed horizontal line indicates the typical neuroepithelial thickness of the mouse PNP .
cultures, the apical area of cells in G2 is significantly larger than those in G1/S in both iPSC lines tested (Fig. 6G). However, M phase cells on curved geometries retain larger apical areas than in G1/S (Fig. 6G). Thus, cell cycle-related regulation of apical area is an evolutionary conserved neuroepithelial behaviour, but mitotic apical constriction is selectively displayed by cells on flat tissue geometries.

Discussion
Cells are often required to multitask, yet their molecular machinery is shared between functions. For example, cytoskeletal remodelling occurs during mitotic rounding and apical constriction, both of which can cooperate to drive morphogenesis (Kondo and Hayashi, 2013). Optimal balance between opposing behaviours is likely to depend on the wider context in which they occur, including the shape of their tissues. Here we show that mitotic apical constriction is a geometry-dependent behaviour which reverses apical expansion in flat neuroepithelia, but its absence prolongs dilatory effects of cell cycle progression in tissues with a convex apical geometry. Tissue geometry, the mechanics of cell packing, and the repertoire of cell behaviours evolve as the neural tube closes. Our findings during neurulation, when regulation of neuroepithelial apical size is a force-generating mechanism, complement studies of later developmental stages during which asymmetrical inheritance of the apical membrane and apical abscission contribute to neurogenesis (Nishizawa et al., 2007;Das and Storey, 2014;Alexandre et al., 2010). It remains to be established whether early morphogenetic cell behaviours pre-emptively achieve a geometry conducive to later differentiation, and whether later functions are regionally diversified by geometric constraints imposed physiologically or pathologically, such as due to failed neural tube closure.
In our previous report we documented that mitotic neuroepithelial cells in the mouse PNP have constricted apices relative to their interphase counterparts through mechanisms sensitive to pharmacological Rho/ ROCK blockade (Butler et al., 2019). Here, we observe the highest apical non-muscle myosin levels in mitotic cells with the smallest apical surfaces, with good correlation between cortical and apicomedial cap myosin pools. This contrasts with invaginating Drosophila mesodermal and ectodermal cells in which apicomedial myosin decreases, and area increases, as cells enter mitosis (Ko et al., 2020). A difference between these Drosophila cells and mouse neuroepithelia is the abundance of cortical myosin in the latter. Cortical myosin predominates in Vangl2-null mouse PNP neuroepithelial cells which apically constrict more than Vangl2-replete cells adjacent to them , suggesting this is the myosin pool primarily responsible for constriction. Consistent with this, junctional shortening is a well-established mechanism of neural plate apical constriction which follows pulsatile myosin accumulation during live imaging in Xenopus embryos (Ossipova et al., 2014).
Live imaging of the chick neuroepithelium shows that cell cycle stage is not the sole determinant of neuroepithelial apical area: the variability in apical dimensions of anaphase cells is larger than the average change during mitotic constriction. The apical surface area of the mouse anterior neuroepithelium decreases between E8.5 and E10.5 (after closure) (Ohmura et al., 2012;Grego-Bessa et al., 2016;Brooks et al., 2020), while retaining equivalent tension inferred from recoil following laser ablation (Bocanegra-Moreno et al., 2022). In the open PNP we observe that neuroepithelial apical areas are smaller in the chick than in the mouse, whereas apical areas of the mouse PNP and human iPSC-derived neuroepithelia are equivalent.
While other cultured epithelial cell types have been productively used to study apical constriction (Yano et al., 2021;Martin et al., 2019), to our knowledge, none are pseudostratified and undergo IKNM in vitro. Here, we characterise the robust differentiation of flat neuroepithelial sheets from human iPSCs through a previously-reported brief and simple dual SMAD inhibition protocol. The molecular identities of cells derived from this protocol have been extensively studied (Chambers et al., 2009;Strano et al., 2020), but we propose a minimum molecular markers panel which, to our knowledge, is unique to neuroepithelia. Of particular interest is this protocol's relatively understudied ability to induce loss of the nuclear Lamin A/C, which facilitates IKNM in vivo (Yanakieva et al., 2019b). Several other iPSC differentiation protocols have been described which produce pseudostratified cells in 3D aggregates (Lee et al., 2022;Karzbrun et al., 2021;Veenvliet et al., 2021), but direct imaging of their apical surface would not be feasible in those systems. The heterogeneity of apical areas within individual cultures is striking, reflecting the in vivo situation despite lacking tissue-level morphology or external secreted cues. We are unable to comment on the dynamicity of their apical surfaces due to these cells' apparent sensitivity to vital dye labelling and confocal/two-photon Z-stack imaging necessary to analyse their apical surface. Nonetheless, the granularity afforded by in situ cell cycle analysis using high-resolution imaging of human-specific Ki-67 (Galea et al., 2013) confirms that dilation occurs in G2.
IKNM nuclear displacement in G2 can be achieved by different molecular mechanisms depending on tissue geometry (Yanakieva et al., 2019a). In the flat zebrafish hindbrain, Rho/ROCK-dependent actomyosin activation rapidly displaces the nucleus apically, whereas in the retina sub-nuclear F-actin accumulation through formin polymerisation displaces the nucleus (Yanakieva et al., 2019a). In a static snapshot, slower ascent would produce a smaller proportion of apical nuclei. We observe more apical nuclei in the convex midbrain than the flat PNP of mouse embryos, consistent with the previous findings (Yanakieva et al., 2019a). Differential rates of nuclear displacement during IKNM has also previously been live imaged in flat versus curved portions of the mouse cochlear epithelium (Ishii et al., 2021). One possible explanation for this is differential expression of actomyosin regulators, either linked to tissue-specific cell identities or secondary to cells' responses to mechanical constraints.
Differential expression of relevant genes has previously been documented. For example, formin homology domain-containing (Fhod)3 is selectively expressed in apically convex regions of the anterior neuroepithelium overlying the rhombomeres and its deletion prevents apical constriction, causing exencephaly in mice (Sulistomo et al., 2019). In Xenopus, simultaneous accumulation of F-actin and N-cadherin (CDH2) parallels apical constriction anteriorly, whereas N-cadherin does not increase in the constricting posterior neuroepithelium (Baldwin et al., 2022). Important differences between Xenopus and amniotes include the lack of a pseudostratified neuroepithelium and brevity of neural tube closure relative to cell cycle progression in the former . Given cell division progressively widens the PNP in the absence of apical constriction in mice (Butler et al., 2019), it is conceivable that synchronisation with cell cycle progression is of greater importance in slower neurulation events. Diminishing proliferation causes exencephaly, but rescues spina bifida in a mouse genetic model (Seller and Perkins, 1983;Copp et al., 1988). Opposition between cell cycle progression and PNP closure is consistent with computational analyses showing that IKNM, modelled without mitotic constriction, promotes apical expansion and nuclear crowding (Ferreira et al., 2019).
Apical expansion relative to the basal domain is characteristic of epithelia with convex geometries, such as the early mouse midbrain studied here. Subtle changes in tissue curvature change the effective force direction produced by cellular contractility (Maniou et al., 2021). Tissue curvature, and the challenges it poses for epithelial cells arrangements, is emerging as a topic of active research (Lou et al., 2022;Prabhakara et al., 2022;G omez-G alvez et al., 2022). Differential cell packing and stress peaks may explain the unexpected differentiation of non-neuroepithelial, squamous cells at the top of glass beads used in these studies. The identity of these CDH2-negative cells is unknown, yet they remain connected to the pseudostratified cells which differentiate along the curved sides of each bead. These cells do show dynamic cell cycle-related changes in apical area: a clear increase between G1/S and G2 as well as constriction between M and G1/S. However, their imposed geometry abolishes mitotic apical constriction. This may be an adaptive process which facilitates expansion of the apical surface relative to the basal, or reflect a limitation of mitotic cells ability to mechanically pull their neighbours' cell junctions over their apical rounded body.
Thus, synchronisation between the cell cycle and apical area is a conserved feature of neuroepithelia, potentially counteracting apical expansion promoted by IKNM (Ferreira et al., 2019). Failure of this synchronisation may impair neural tube closure by causing progressive widening of the neuropore as cells divide (Butler et al., 2019). This synchronisation is conserved in a tractable and validated human iPSC-derived model. The ability to empirically change these cells' geometry corroborates in vivo observations that mitotic apical constriction does not happen in apically convex neuroepithelia, retaining larger apical areas until they return to the G1 phase. Persistent dilation during M phase correlates with the expanded apical relative to basal surface in convex epithelia. These findings suggest neuroepithelial cells are not only able to multitask by balancing growth through proliferation with neuropore shrinkage through apical constriction, but do so while interpreting the physical cues of their tissue structure.

Mouse embryo collection
Studies were performed under project and personal licenses regulated by the UK Animals (Scientific Procedures) Act 1986 and the Medical Research Council's Responsibility in the Use of Animals for Medical Research. Pregnant female mice were time-mated and sacrificed by cervical dislocation 8.5 or 9.5 days after the morning a plug was identified. C57Bl/6J mice were bred in-house and mated when at least six weeks old. mTmG reporter mice used to visualize cell outlines (Muzumdar et al., 2007), Nkx1.2 CreERT2 used to mosaically label PNP cells (Rodrigo Albors et al., 2018), Sox2 CreERT2 used to mosaically label anterior neuroepithelial cells (Andoniadou et al., 2013) bred in house, and their administration of oral tamoxifen (50 μl of 100 mg/ml solution in corn oil per mouse), were all as previously described (Savery et al., 2020;Galea et al., 2021).

Chick embryo culture and live imaging
Fertilised Dekalb white chicken eggs (Henry Stewart, Norfolk, UK) were incubated at 37 C in a humidified chamber for~36 h reaching Hamburger and Hamilton stages 9-10. Embryo collection and culture was according to the EC protocol (Chapman et al., 2001) with the slight modification of utilising a double filter paper sandwich as a carrier. Any excess yolk was washed off with Pannett-Compton saline. Vitelline membrane windowing was done with a tungsten needle.
Embryos were first stained with a 1:1 mix of 1:50 Hoechst (Ther-moFisher Scientific #62249) and 3:10,000 ice-cold DiI suspension (Invitrogen #D282) in PBS for 15 min at 37 C. Following washing with Pannett-Compton saline, embryos were further stained with 1:100 Cell-Mask Deep Red (Invitrogen #C10046) plasma membrane stain in PBS at 37 C for 15 min. Any excess stain was washed off with Pannett-Compton saline. Live imaging was done with a Zeiss Examiner LSM880 confocal under AiryScan Fast mode using a 20x/NA0.7 C-Epiplan Apochromat dry objective. Hoechst was excited using a MaiTai tuneable two-photon laser to avoid phototoxicity.
The protocol for direct differentiation of hiPSCs into neuroepithelial sheets was as previously described (Shi et al., 2012a), with minor modifications. Cells differentiated following this protocol can form neurons with a dorsal forebrain identity (Shi et al., 2012a). Briefly, cells were lifted using EDTA solution and plated in 1% Matrigel®-coated tissue culture plates. Approximately~280,000 cells/cm 2 were transferred into a 35 mm glass bottom dish (ibidi, Thistle Scientific) ensuring 100% confluency the following day. After 24 h, the cells were washed twice with PBS and medium was replaced with neural induction medium consisting of an equal ratio of N2 (N2 supplement in DMEM/F12, Thermo Fisher 17502048 and 31331093) and B27 solution (B27 supplement in Neurobasal, Thermo Fisher 17504044 and 12348017) supplemented with 10 μM SB431542 and 1 μM Dorsomorphin (both Tocris Bioscience).
The medium was changed daily, and cells were cultured for up to eight days.

iPSC differentiation on glass beads
150-212 μm diameter glass beads (Sigma G1145-10G) were used to generate curved epithelial geometries, corresponding to the range of basal midbrain curvatures observed in early mouse embryos. Beads were autoclaved and transferred using a pipette tip onto the centre of a 35 mm glass bottom dish. The bead and dish glass were coated with 1% Matri-gel® GFR for 1 h at room temperature. iPSCs were then seeded and neuroepithelial differentiation performed as above.
Images were processed and visualized as 3D or maximum projections in Fiji (Schindelin et al., 2012). PureDenoise and 5-pass Richardson Lucy deconvolution in Deconvolution Lab (Sage et al., 2017) were applied to live-imaged chick datasets (distributed by the Biomedical Imaging Group, EPFL). The elevation map was generated using the Fiji temporal LUT, replacing T with Z. Surface subtraction was performed essentially as previously described (Galea et al., 2018) using an in house Fiji macro available from https://github.com/DaleMoulding/Fiji-Macros (courtesy of Dr Dale Moulding).
Cell apical areas were manually segmented in 3D confocal Z-stacks, following the cell's outline from its nucleus to its apical surface. The actomyosin cortex is readily identifiable as a bright rim around each cell's apical surface. The inner border of this cortex was used to define an area selection representing the apical cap, in which myosin intensity was quantified. The perimeter of this area selection was then converted into a line and expanded away from the cell's centre by 0.5 μm. The resulting band was used to define the apical cortex. Thicknesses, intensity, length, and areas were measured using in-built Fiji functions. In each case, every mitotic cell in each field of view was analysed and at least as many nonmitotic cells were analysed in the same field of view, analysing contiguous non-mitotic cells whenever possible.
Mosaic labelling of neuroepithelial cells was achieved with Bio-Tracker™ 490 Green Cytoplasmic Membrane Dye (Thermo Fisher Scientific). Fixed cells were stained according to the manufacturer's instructions for 30 min at room temperature.

Western blotting
Western blotting was performed essentially as previously described (Thompson et al., 2022;Galea et al., 2020). Cells were lysed on ice for 30 min in 50 mm Tris Base pH 7.6, 150 mm NaCL, 1% Triton X-100, 0.02% Sodium Azide, 1 mm protease inhibitor cocktail (Merck Life Science UK Ltd), 1 mm sodium orthovanadate, 25 mm sodium fluoride. Protein concentration was determined using Pierce BCA Protein Assay kits (Thermo Fisher). Proteins were resolved by SDS-PAGE using 10% polyacrylamide gels and transferred to PVDF membranes. Membranes were blocked in TBS (15.4 mm Trizma-HCL, 4.62 mm Tris-base, 150 mm NaCl, pH 7.6) containing 10% w/v milk powder (Merck Life Science UK Ltd). Membranes were incubated with primary antibody overnight and then with horseradish peroxidase (HRP)-conjugated secondary antibodies (Agilent Technologies, Stockport, UK) for 1 h. Primary antibodies were as described in the immunofluorescence section. HRP was detected using ECL Prime (Cytiva, Amersham, UK). To strip and re-probe, membranes were incubated in 0.2 m NaOH for 20 min at 37 C and then 20 min at room temperature before re-blocking and re-use.

Statistical analysis
Comparison of two groups was by two-tailed t-test, paired when measures were repeated (e.g. same cell at two time points). Comparison of more than two groups was by one-way or two-way ANOVA with posthoc Bonferroni correction. Linear regression was by Pearson's correlations. Sine curve fitting and all statistical tests were performed in Origin 2021. P < 0.05 was considered statistically significant. Blinding was not possible due to obvious differences in tissue geometry or the need to identify individual cells.
Wherever possible, the embryo or independent culture dish (cultured on different days) was considered the unit of measure. When this was not possible, e.g. due to small numbers of mitotic cells in each not providing a representative value, the cell was considered the unit of measure. All mitotic cells fully present in an imaged field of view were included in each analysis. All representative data is based on at least four independent (different culture/litter) observations.

Author contributions
GLG designed the study in discussion with EM, PDC, NE, AJC, NDEG and FL. IA performed all iPSC work and analysis with help from EMT, TW, FP, GGG, and CM. GLG performed all mouse work and analysis with help from CE and EM. Chick embryo work and analysis was performed by CE with help from EM and GLG. GLG, EM and PDC supervised students. GLG, PDC, NE, AJC, NDEG, FL, CM, GGG and FP provided resources. GLG drafted the manuscript, and all authors approve the version submitted.

Declaration of competing interest
The authors declare they have no conflicts of interest.

Data availability
Data will be made available on request.