Scribble and E-cadherin cooperate to control symmetric daughter cell positioning by multiple mechanisms

The position of the two daughter cells following cell division is critical to many biological processes, from embryonic patterning and tissue growth to the extrusion of mutant cells (Biggins et al., 2015; Dekoninck et al., 2020; Lechler and Mapelli, 2021). Daughter cell positioning is coordinated by a complex interplay of responses to intrinsic and extrinsic cues. These cues can include receptors at cell–cell contact sites and the extracellular matrix, and non-chemical influences, such as cell geometry and tissue tension (van Leen et al., 2020). Cues are translated to daughter cell positioning through cell activities such as remodelling of junctions and orienting of the mitotic spindle (Gillies and Cabernard, 2011). An active question in the field is how these cues enable memory of the original position of the parent, and are transmitted to control position and fate of the daughter cells.

Spindle orientation provides a functional link between spatial context and fate of the progeny of a cell division in many contexts. For instance, spindle orientation influences the position, the size and the fate of the two daughter cells of an epithelial division, impacting upon cell diversification and tissue homeostasis and morphogenesis (Bergstralh et al., 2017; Dewey et al., 2015; di Pietro et al., 2016; Lu and Johnston, 2013; Seldin and Macara, 2017). Errors in the control of orientation of the mitotic spindle lead to developmental defects and cancer (Bergstralh and St Johnston, 2014; Lechler and Mapelli, 2021; Lu and Johnston, 2013). Recent studies have indicated that several attributes combine to influence spindle orientation and daughter cell positioning, including intrinsic polarity, the location of adhesions and constraints on the cell shape (Charnley et al., 2013; Dimitracopoulos et al., 2020; Lesman et al., 2014; Li et al., 2019; Li and Burridge, 2019; Matsumura et al., 2016; Mitchison, 1992; Nestor-Bergmann et al., 2019; Niwayama et al., 2019; Petridou and Skourides, 2016; Rizzelli et al., 2020; Thery and Bornens, 2006). All these cues are transmitted to spindle orientation via positioning of the LGN (also known as GPSM2) complex for a final spindle orientation (Lechler and Mapelli, 2021).

Epithelial cells physically interact with the surrounding extracellular matrix via integrins and with neighbouring cells via adherens junctions, providing several possible such molecular means of transmitting cues for spindle orientation (Walma and Yamada, 2020). As the cell rounds up in metaphase, protrusions termed retraction fibres can maintain connection to the substrate (Anastasiou et al., 2020; Finegan and Bergstralh, 2019; Fink et al., 2011; Lam et al., 2020; Nunes and Ferreira, 2021; Petridou and Skourides, 2016; Thery et al., 2005). Characterisation of these fibres suggests a context-dependent molecular composition at the site of tethering to the substratum, which then propagates tensile forces through the fibre (Kotak and Gonczy, 2013; Nestor-Bergmann et al., 2014). The molecules that provide tensile strength to the retraction fibre, and mediate the functional connection from fibre to LGN, are not yet well understood.

Subsequent to spindle orientation, membrane remodelling provides a second critical step that ensures that daughter cells are appropriately positioned following cell division (Gibson et al., 2006). The mechanisms by which daughter cells reclaim the space vacated by their parent involve a similarly complex interplay involving adhesion complexes, tension and geometry (Pinheiro and Bellaiche, 2018). In epithelial monolayers, this process is complicated by the need to maintain epithelial barrier function, which necessitates continual interaction with neighbouring cells (Guillot and Lecuit, 2013). As with spindle orientation, membrane remodelling involves a dynamic relationship between the dividing cell and its neighbours, with physical forces and signalling adhesion proteins, such as E-cadherin (Bui et al., 2016; Guillot and Lecuit, 2013) and the cell-competition mediator Scribble (Maruyama and Fujita, 2022; Ogawa et al., 2021). The mechanisms of membrane remodelling in the absence of neighbouring cells has been less well studied.

Here, we used a simple model of single MCF10A cells dividing on glass to explore the minimal requirements for spindle orientation, membrane remodelling and daughter cell positioning. We made the surprising finding that E-cadherin and Scribble dictate spindle orientation even without neighbouring cells. E-cadherin and Scribble subsequently relocate to the nascent junction between the two daughter cells, and orchestrate a spreading of that junction to facilitate enduring connection between the daughters. The findings described below show that Scribble traffics dynamically between key regions of the cell during cell division, coordinating the E-cadherin-mediated effects on the spindle and the nascent daughter–daughter junction.

We first explored the molecular composition of cell protrusions during division of single MCF10A human mammary epithelial cells. During prometaphase, both the cell body and retraction fibre were rich in F-actin and E-cadherin, with low levels of active integrin β1 (Fig. 1Ai). By the telophase stage, the cell contained shorter protrusions, with more active integrin β1 in the adhesions and less E-cadherin (Fig. 1Aii, quantified in Fig. 1Aiii). The enrichment of E-cadherin in prometaphase retraction fibres was a surprise given that E-cadherin is classically considered to mediate cell–cell interactions, rather than cell–matrix interactions (Lecuit and Yap, 2015; Pannekoek et al., 2019). It also does not seem likely that E-cadherin in the retraction fibres is important for adhesion or recruited in a ligand-dependent manner, given that the substrate was not coated with E-cadherin ligands and that the E-cadherin signal in the fibres was no stronger at the contacts of between fibres and the substrate than it was along the fibre lengths (see xy view of E-cadherin, Fig. 1A). Further characterisation indicated that the integrin β1 -associated protein, paxillin, was also not detected in metaphase protrusions, but was clearly at the base of the protrusions by late telophase (Fig. S1A). Of proteins known to mediate E-cadherin functions, β-catenin, but not β-PIX (also known as ARHGEF7), was detected in protrusions throughout mitosis (Fig. S1B,C). Together, these findings indicate that the retraction fibres of single MCF10A cells early in mitosis contain E-cadherin, and at later stages, E-cadherin is mostly replaced by integrin β1. Whether the tips of the fibres rely on integrins for adhesion (for instance β1 or β5), and whether the context (epithelial or not, uncoated or ECM versus glass) determines the composition of the fibres is not clear (Dix et al., 2018; Lock et al., 2018).

Given the role of retraction fibres in positioning of daughter cells after cell division (Taneja et al., 2019), we proposed that E-cadherin enrichment in retraction fibres during mitosis might indicate a role for E-cadherin in daughter cell positioning. We assessed this using a blocking antibody to E-cadherin, HECD-1 that prevents E-cadherin–E-cadherin interactions (Shimoyama et al., 1989; Tomlinson et al., 2001) (Fig. 1B; Movie 1). We first explored the time taken for the daughter cells to re-adhere and spread on the substrate after cell division. In untreated cells, both daughter cells adhered to the substrate within a relatively short time (Fig. S2A). However, in the presence of HECD-1 treatment, we consistently observed that of two daughter cells produced per division, one was significantly inhibited in its re-adherence and re-spreading, while the other was unaffected. This was not due to diversion of E-cadherin trafficking by the HECD-1 antibody, as co-staining after treatment showed colocalisation of HECD-1 and anti-E-cadherin (Fig. S2B). Thus, E-cadherin is not required for spreading of the daughter cells per se during cytokinesis, but is required for optimal positioning and re-adherence of the second daughter cell after division of MCF10A cells. The distribution of E-cadherin along the retraction fibres, and the normal adhesion of one daughter under E-cadherin inhibition suggests that, in this context, E-cadherin has a function other than matrix adherence.

One possible explanation for these findings, particularly given that E-cadherin enrichment in the retraction fibres is most evident early in mitosis, might be that E-cadherin is required for the mitotic spindle to be oriented parallel to the substrate during division, ensuring both daughters have access to substrate for re-adherence following mitosis. Retraction fibres are known to dictate spindle orientation, although integrins rather than E-cadherin have been implicated in this function (Fink et al., 2011; Taneja et al., 2019). On the other hand, E-cadherin can regulate spindle orientation in the context of an epithelial tissue (den Elzen et al., 2009; Gloerich et al., 2017; Hart et al., 2017; Wang et al., 2018a), and it has been recently shown that E-cadherin knockdown disrupts spindle orientation in the immortalised prostate epithelial cell line RWPE-1 (Wang et al., 2018a). In some contexts, spindles are randomly oriented in metaphase, and only achieve appropriate orientation in anaphase, but in other contexts, alignment occurs in metaphase (di Pietro et al., 2016; Kotak, 2019; Machicoane et al., 2014; Yamashita et al., 2003). Single MCF10A cells showed the latter behaviour, with spindles reliably aligned parallel to the substrate in metaphase, anaphase and telophase (Fig. 1C). However, treatment with HECD-1 (Fig. S2C) or depletion of E-cadherin using siRNA (si-Ecad) (Fig. 1C; Fig. S2D), demonstrated that disruption of E-cadherin caused orientation of the spindle away from the substrate. Thus, inhibiting E-cadherin leads to asymmetric adhesion of the daughter cells, and a mis-oriented spindle in the dividing mother cell, similar to previously observed findings of integrin β1 in cultured HeLa cells.

Together, these findings suggest that E-cadherin plays an unexpected role in single dividing cells, influencing daughter cell positioning and controlling the orientation of the mitotic spindle. The fact that these cells have no neighbours, and that E-cadherin is enriched in retraction fibres, suggests that, like integrins, E-cadherin might transmit information to the spindle via retraction fibres, perhaps in cooperation with adhesions mediated by integrin β1 (Toyoshima and Nishida, 2007) or integrin β5 (Lock et al., 2018).

It is well established that stabilisation of E-cadherin is key to its function at cell junctions, where E-cadherin stability is promoted by homophilic adhesions (Cavey et al., 2008). To learn what might promote stability of E-cadherin at the cell–substrate interface, we explored the localisation of the polarity protein, Scribble. Scribble and E-cad reciprocally regulate the positioning and stabilisation of each other in a context-dependent manner, which is compatible with the notion that transient structures such as those formed during mitosis might particularly feature a co-dependent recruitment of Scribble and E-cadherin (Allam et al., 2018; Bonello and Peifer, 2019; Choi et al., 2019; Navarro et al., 2005; Qin et al., 2005). Staining of MCF10A cells undergoing mitosis showed Scribble, E-cadherin and F-actin at the cortex and in irregular foci along retraction fibres (Fig. 2A). xz views of the dividing cells indicated that cortical Scribble, E-cadherin and F-actin were consistently enriched at the poles of the cells in metaphase (Fig. 2B). This position is consistent with the previously observed localisation of active integrin β1 to retraction fibres at their convergence at cortical poles, where it recruits the LGN complex to control spindle orientation in single mitotic HeLa cells (Petridou and Skourides, 2016).

To test whether Scribble impacted E-cadherin stability in either retraction fibres or the cortex, we depleted Scribble in MCF10A cells using two knockdown approaches: siRNA and shRNA. Scribble was uniformly depleted by both knockdown approaches (Fig. S3A). Depletion of Scribble had no impact on adhesion following cell trypsinisation (Fig. S3B), and no impact on either total or cortical levels of E-cadherin at the cortex of cells in a confluent monolayer (Fig. S3C). However, single Scribble-depleted cells showed almost complete loss of E-cadherin in retraction fibres at prometaphase (Fig. 2C; Fig. S3D) and at the cortex at metaphase (Fig. 2D; Fig. S3E). This, combined with the normal expression of E-cadherin in Scribble-depleted MCF10A cells in confluent interphase cells, suggests that Scribble is required for the recruitment of E-cadherin in dynamic structures at the cortex and retraction fibres during cell division. These findings are consistent with the notion that context alters the phenotype of Scribble depletion, with more obvious phenotypes in dynamic situations with less opportunity for compensation (Allam et al., 2018; Bonello and Peifer, 2019)

All three members of the Scribble complex [Scribble, Discs large (Dlg; Dlg1–Dlg5 in humans) and Lethal Giant Larvae (LGL; LLGL1 and LLGL2 in humans)] mediate spindle orientation in intact tissues, with this role confirmed in vertebrates for Dlg1 and Scribble (Allam et al., 2018; Bell et al., 2015; Bonello and Peifer, 2019; Carvalho et al., 2015; Godde et al., 2014; Nakajima et al., 2019; Nakajima et al., 2013; Porter et al., 2019; Qin et al., 2005; Wang et al., 2018a; Zigman et al., 2011), so we next investigated whether Scribble is required for spindle orientation in mitotic single cells. We assessed spindle orientation in MCF10A cells depleted of Scribble using shRNA (sh-Scrib) or siRNA. In Scribble-depleted cells undergoing cell division, the spindle was mis-oriented (Fig. 3A; Fig. S3F). This phenotype was similar to that observed with loss of function of E-cadherin (Fig. 1C). Thus, both Scribble and E-cadherin regulate spindle positioning.

To test our hypothesis of information flow from E-cadherin and Scribble, perhaps through mediating spindle orientation, to effects on daughter cell positioning, we first tracked Scribble-depleted cells using time-lapse microscopy (Fig. 3B; Movie 2). Depletion of Scribble led to a slight increase in the time taken for daughter 1 to re-adhere to the substrate, and a dramatic increase for daughter 2. Indeed, some of the daughter 2 cells failed to re-adhere before they were lost to the tracking, perhaps because the spindle mis-orientation reduced association with the substrate. Therefore, Scribble is required for daughter cell positioning, with an asymmetric adherence of one daughter, as we had previously seen with E-cadherin inhibition. In confluent monolayers, E-cadherin is stabilised at regions of high tension, to which it directly recruits LGN (Gloerich et al., 2017; Hart et al., 2017). LGN then engages with NuMA (also known as NUMA1) upon breakdown of the nuclear envelope, which orchestrates pulling on astral microtubules. To assess whether a similar process occurred in single cells, we depleted cells of the canonical spindle regulator LGN (sh-LGN). Depletion of LGN disrupted the spindle orientation (Fig. 3C). Remarkably, depletion of LGN also caused asymmetric adhesion of the daughter cells, with a delay in adherence of the second daughter, although with less of a delay than seen with E-cadherin inhibition (Fig. 3D; Movie 3). Our data are compatible with recent findings showing E-cadherin in single mitotic cells controlled positioning of astral microtubules with the complex involving EB1 (also known as MAPRE1) and NuMA for spindle positioning (Gloerich et al., 2017; Song et al., 2021). These data together suggest a signalling pathway from E-cadherin, through Scribble and LGN–NuMA, to spindle orientation, and indicate that the asymmetric daughter cell adherence is, at least in part, due to disruption of the spindle orientation.

We next explored whether Scribble and E-cadherin utilised the canonical mediators of spindle orientation (di Pietro et al., 2016). Spindle orientation is dictated by dynein-mediated forces, which are generated when the microtubule motor protein dynein pulls on astral microtubules emanating from the spindle pole (Kotak, 2019; Thery et al., 2007). The position of the dynein-based motor complex is controlled by the position at the cortex of a complex termed the LGN complex [comprising Gαi (also known as GNAI1), LGN and NuMA in mammals; di Pietro et al., 2016; Kiyomitsu, 2019]. The positioning of the LGN complex is highly context specific, but is often controlled by polarity proteins, so we speculated that Scribble might function by enabling the E-cadherin-based recruitment of the LGN complex to cortical poles. In wild-type single MCF10A cells, NuMA localised to spindle poles and the cortex at metaphase and anaphase, but was not present in the retraction fibres (Fig. 4A,B). NuMA was not evident at the telophase cortex (Fig. S4A). This change in NuMA localisation is compatible with the previously described cell cycle-dependent influence of LGN on NuMA localisation (Du and Macara, 2004; Zheng et al., 2014). To determine any functional relationship between Scribble and the LGN complex, we assessed NuMA localisation during metaphase in the context of Scribble depletion. Both methods of Scribble depletion reduced the cortical recruitment of NuMA at metaphase (Fig. 4C; Fig. S4B). The specificity of this effect is demonstrated by our observation that Scribble depletion did not impact the localisation of myosin IIb (Fig. S4C). Whether the effect of Scribble on NuMA localisation is direct, or via its effects on E-cadherin, is not known.

Scribble played similar roles in anaphase as we had observed in metaphase cells. As previously reported (Zheng et al., 2014), NuMA levels are dramatically increased at anaphase (Fig. 4B), and, similar to what is seen in metaphase, at anaphase, E-cadherin and Scribble colocalised to the cortical pole (Fig. 4D). Again, depletion of Scribble using either siRNA or shRNA prevented cortical localisation of E-cadherin at anaphase (Fig. 4E; Fig. S4D). Similarly, depletion of Scribble reduced cortical NuMA at anaphase (Fig. 4F; Fig. S4E). These results suggest that cues from the substrate are transmitted through E-cadherin and Scribble, perhaps via retraction fibres, to recruit NuMA to the cortical pole at the base of the fibre for spindle orientation.

Dlg, a key functional partner of Scribble, can colocalise with Scribble at the cortical pole, where Dlg recruits the LGN complex to orient the mitotic spindle in several animal models and tissue types (Allam et al., 2018; Bell et al., 2015; Bergstralh et al., 2013; Johnston et al., 2009; Nakajima et al., 2019; Nakajima et al., 2013; Saadaoui et al., 2014). To assess this, we stained for Dlg in control and Scribble-depleted dividing cells using antibody that reacts with Dlg1–Dlg4 (Fig. 5A; Fig. S4F). Like Scribble, Dlg was cortical throughout mitosis and clearly present in retraction fibres. Depletion of Scribble clearly reduced cortical Dlg in mitosis. To assess whether this reflected alterations in localisation or expression levels, we fixed and permeabilised cells and performed flow cytometry. Importantly, Scribble depletion had a negligible effect on both Dlg and E-cadherin level in unsynchronised (predominantly interphase) cells (Fig. 5B). Together, these results suggest that E-cadherin and Scribble coordinate with NuMA and perhaps Dlg at the two cortical poles to ensure stable alignment of the spindle parallel to the substrate.

In addition to the role of spindle orientation, we identified a second possible mechanism for coordinating daughter cell adherence and spread. In the context of a confluent monolayer where mitotic cells were surrounded by neighbouring cells, we observed that, in late telophase, Scribble relocated to the region between the two nascent daughters (Fig. 6A). Scribble was not evident at the basal adhesion region between the two daughters where both F-actin and paxillin were enriched, but was strongly expressed at the daughter–daughter contact site above the intracellular bridge (Fededa and Gerlich, 2012; Rathbun et al., 2020) (Fig. 6A). A similar relocation occurred in single dividing cells, whereby Scribble, F-actin and myosin IIb were broadly recruited to regions of furrow ingression and daughter–daughter contact sites at progressive telophase stages (Fig. S5), and Scribble was co-enriched with F-actin at the daughter–daughter contact rather than with myosin IIb. However, activity of both actin and myosin was required to promote Scribble recruitment or stability, as treatment with Cytochalasin D and Blebbistatin abrogated Scribble at the daughter–daughter interface (Fig. S6A,B). E-cadherin was co-enriched with Scribble at the nascent daughter–daughter junction (Fig. 6B). Dlg, which has previously been found at nascent daughter–daughter contacts (Li et al., 2018), was also recruited to the daughter–daughter contact site in late telophase cells (Fig. 6C), and interestingly, Dlg, but not Scribble, was enriched at the midbody.

To assess the molecular control of recruitment to the daughter–daughter contact, we first assessed how the absence of Scribble changed the composition of the nascent junctions. Scribble was required for optimal recruitment of E-cadherin to the interface during cytokinesis [cells were analysed at a similar stage, as indicated by the myosin IIb staining pattern (Fig. S5)] (Lecuit and Yap, 2015; Wang et al., 2019) (Fig. 7A). This dependency on Scribble was less striking for E-cadherin in the myosin II-enriched adherens junction of interphase cells within a confluent monolayer. Depletion of Scribble also reduced the recruitment of Dlg at the daughter–daughter interface (Fig. 7B; Fig. S7A).

A caveat to the analysis of the molecular control of recruitment to the nascent interface was our observations that the nascent junction seemed to be smaller when Scribble was depleted. This observation also suggested that, as has been observed in E-cadherin-based adhesion in interphase MDCK cells (Li et al., 2020; Papalazarou and Machesky, 2021), E-cadherin and Scribble might coordinate a membrane reorganisation at the nascent daughter–daughter contacts. This nascent junction reorganisation might coordinate subsequent positioning of the two daughters. We therefore explored the functional relationship between the extent (‘width’) of the interface, molecular recruitment to the interface and the shortest distance between the DNA (a surrogate for cell positioning) of each daughter after Scribble depletion. Compatible with the findings in Fig. 6, Scribble levels at the nascent junction were well correlated with E-cadherin levels at the nascent junction, indicating that Scribble recruitment is upstream of E-cadherin recruitment (Fig. 7C; Fig. S7B). A strong correlation between the width of the contact and the levels of E-cadherin further raised the possibility that the effect of Scribble on E-cadherin at the nascent junction might be mediated by the effect of Scribble on the extent of the daughter–daughter interface. Together, these findings indicate that positioning of Scribble at telophase controls the width of the nascent daughter–daughter contact, with a subsequent effect on recruitment of other molecules such as E-cadherin and Dlg.

Having shown that depletion of Scribble reduced the width of the nascent junction, we assessed how this interface correlated with the positioning of the two daughter cells. With depletion of Scribble, the reduction in width of contact was not well correlated with distancing of the two nuclei from each other (Fig. 7C; Fig. S7B), suggesting that the extent of the contact was not important for early (at telophase) positioning of the two daughter cells. However, it should be noted that only cells in which both daughters adhered could be analysed for nuclear distance, rendering the Scribble data set skewed. These data suggest the possibility of a causal relationship between nascent interface composition and the subsequent positioning of the daughters.

Taken together, the effect of Scribble and E-cadherin on daughter cell positioning could be mediated through a combination of spindle orientation and the integrity of the nascent daughter–daughter junction.

By exploring MCF10A cells without neighbours, we have identified a novel cell-autonomous mechanism of controlling spindle orientation and daughter cell positioning. Our experiments reveal that in the absence of neighbouring cells, E-cadherin can transmit signals from the substrate to NuMA, a component of the LGN complex. In contrast to the situation during interphase, the stability and position of E-cadherin is profoundly dependent on Scribble during cell division. We also show that remodelling during cell division leads to further functions for E-cadherin and Scribble in the expansion of a nascent daughter–daughter junction, which controls the positioning of both daughters at and beyond telophase (Fig. 8A,B).

E-cadherin is well known to orient the mitotic spindle by transmitting signals from cell neighbours via adherens junctions (Gloerich et al., 2017; Hart et al., 2017). Similarly, integrins can orient the mitotic spindle by transmitting signals from the cell matrix via retraction fibres (Li and Burridge, 2019). In single MCF10A cells without an artificial matrix, we were surprised to find E-cadherin, in preference to integrin β1, in the retraction fibres (although integrin-β1 was upregulated by telophase). This finding adds to the growing appreciation that multiple receptors can input signals from the extracellular matrix to control mitotic spindle orientation, including integrin β1 in cells plated on fibronectin (Dix et al., 2018) and integrin β5 in reticular adhesions (Lock et al., 2018).

Provision of artificial ligand on the adhesion surface has previously been shown to orchestrate spindle orientation (Charnley et al., 2012; den Elzen et al., 2009; Petridou and Skourides, 2016), but in our system, there is no apparent ligand available, suggesting a ligand-independent role for E-cadherin. Indeed, coating of the substrate with recombinant E-cadherin as a homophilic ligand on the basal surface of cells has previously been reported to lead to spindles being oriented away from the substrate (den Elzen et al., 2009). A precedent for ligand-independent orienting of the spindle is that mediated by integrin β1 (Petridou and Skourides, 2016). In support of a similar ligand-independent role for E-cadherin, cis-clustering (also known as lateral clustering) of E-cadherin can transmit signals under conformational constraints associated with membrane immobilisation (Thompson et al., 2020; Thompson et al., 2019; Thompson et al., 2021). Cells use cadherin-based tension sensing to monitor and respond to the stiffness of their environment (Discher et al., 2005; Pannekoek et al., 2019; Slováková et al., 2022; Uroz et al., 2018), and E-cadherin dilution can mediate force transduction (Pinheiro and Bellaiche, 2018). Together with our findings, these observations suggest that the recruitment of E-cadherin and Scribble allows for transmission of information from the underlying substrate to orchestrate the spindle. We further suggest that this information might be mediated by retraction fibres, and might involve sensing of tension, perhaps via stretching of the retraction fibres.

The role of Scribble in spindle orientation is well established (Bonello and Peifer, 2019; Godde et al., 2014; Nakajima et al., 2013; Qin et al., 2005; Wang et al., 2018a). However, the complexity of the tissues in which this role has been investigated means that it is still not clear what extrinsic or intrinsic cues are transmitted by Scribble (Godde et al., 2014; Nakajima et al., 2013). Here, we show that Scribble can directly link positional information from E-cadherin at the substrate, where it is concentrated in retraction fibres and the cortical poles to control the spindle orientation. Our finding that Scribble depletion leads to reduction of E-cadherin in the cortex (and retraction fibres) of mitotic, but not interphase, MCF10A cells suggests that mitosis triggers a new form of regulation of E-cadherin by Scribble. Whether this regulation relates to the previously identified impact of Scribble on E-cadherin endocytosis (Lohia et al., 2012) or other mechanisms remains to be determined. Many questions remain, but these observations together support a new model by which Scribble and E-cadherin can orchestrate spindle pole orientation independently of neighbouring cells.

Despite the recognised importance of positioning of the daughter cells following mitosis for tissue organisation during development, metastasis and other pathologies, the means by which adhesive contacts are remodelled following cell division are still unclear (Osswald and Morais-de-Sa, 2019; Pérez-González et al., 2022). This understanding is particularly complicated by the need for cells in a monolayer to disengage adherens junctions with neighbouring cells alongside formation of a new adhesive interface (Le Bras and Le Borgne, 2014; Zulueta-Coarasa and Rosenblatt, 2022). As with the spindle orientation, exploring single cells has allowed us to identify novel mechanisms by which the daughter–daughter contact is established, although our findings suggest these mechanisms are also conserved in an intact monolayer. We find that Scribble, E-cadherin and Dlg are recruited to the intracellular bridge that remains in late telophase, where they mediate an expansion of the adhesive interface between the two daughter cells.

Interestingly, tricellular junctions can contain Scribble, Dlg and E-cadherin (Bosveld and Bellaiche, 2020; Sharifkhodaei et al., 2019), and Dlg in tricellular junctions of the Drosophila dorsal thorax epithelium mediates disentanglement of daughter and neighbour cells, and formation of the nascent daughter–daughter interface (Wang et al., 2018b). Our findings are also reminiscent of a previously identified role of E-cadherin, Arp2/3 and myosin in the formation of a nascent junction in Drosophila dorsal thorax epithelial cells, although, in that context, neighbouring cells were required to achieve a long interface (Herszterg et al., 2013). Given the functional association of Rac1 and Scribble in membrane protrusions and cell extrusion (Dow et al., 2007; Humbert et al., 2006; Pérez-González et al., 2022), it is tempting to speculate that Scribble might also be required in this context.

Together, these studies identify cell-autonomous roles for Scribble and E-cadherin at several stages in the remodelling of spindle and cell membranes during cell division of a single MCF10A cell. These roles were previously obscured, presumably because of the importance of Scribble and E-cadherin in the junctions of epithelial cells in monolayers. Scribble connects E-cadherin (and perhaps other cues) to spindle orientation machinery, cell protrusions and the nascent daughter–daughter contact. This dynamic repositioning of Scribble, and the co-dependence of E-cadherin and Scribble, when the cell is undergoing rapid shape change has precedence in other biological situations (Allam et al., 2018; Bonello and Peifer, 2019; Ludford-Menting et al., 2005; Osmani et al., 2006). Thus, we propose a dynamic role for Scribble throughout cell division in orchestrating the placement of daughter cells.

MCF10A cells were sourced from ATCC, routinely STR profiled and tested for mycoplasma infection, and cultured in Dulbecco's modified Eagle's medium with F12 (DMEM:F12; Gibco), supplemented with 5% (v/v) horse serum (Gibco), 10 μg/ml insulin (Novo Nordisk), 0.5 μg/ml hydrocortisone (Sigma-Aldrich), 20 ng/ml human epidermal growth factor (Sigma-Aldrich), 100 ng/ml cholera toxin (List Biological Labs), 100 ng/ml penicillin / streptomycin, 2 mM glutamine, and maintained at 37°C in 5% CO₂. Stable MCF10A cell lines that expressed short hairpin RNA against human Scribble (#7) and the control were created and analysed as described in Dow et al. (2007). Scribble depletion was re-confirmed by flow cytometry (Fig. S2A). Stable MCF10A cell lines that expressed short hairpin RNA against LGN and the control (sh-Ctrl) were created and examined by immunofluorescence (Fig. 3C). The sequence of human LGN shRNA was 5′-CCAGAGGGCTAGTTTCAGTAA-3′; the sequence of control shRNA was 5′-TCTCGCTTGGGCGAGAGTAA-3′.

Protein level of Scribble, E-cadherin and Dlg were assessed in unsynchronised and confluent MCF10A cells. To collect the cells from culture dish, 2.5 mM EDTA was applied to the culture medium for 20 min in incubator before trypsinisation. Then, the cells were fixed with IC Fixation Buffer (eBioscience) according to manufacturer's protocol, followed by incubation with primary antibody against Scribble, E-cadherin and Dlg, respectively (E-cadherin, Cell Signaling Technology; cat. no. 3195, dilution 1:100; Scribble, Santa Cruz Biotechology, cat. no. sc-11049, dilution 1:250; Dlg, NIH NeuroMab, cat. no. 75-029; dilution 1:500). After labelling with Alexa Fluor-conjugated secondary antibody, the cells were examined using flow cytometry (BD FACSAria III, BD).

Scribble and E-cadherin were depleted by siGENOME SMARTpool siRNA against human Scribble and E-cadherin, respectively [Dharmacon ON-TARGETplus siRNA SMARTPool; control non-targeting pool (OTP-NT) cat. #D-001810-10-20; Scribble (SCRIB, Human), cat. #L-010500-00-0010; E-cadherin (CDH1, Human), cat. #L-003877-00-0010]. siRNA was transfected by using DharmaFECT 3 according to the manufacturer's protocol, and was incubated with cells in serum-free cell culture medium for transfection. We applied two rounds of siRNA silencing to increase knockdown efficiency. Briefly, the first round of siRNA silencing (20 nM of siRNA) was applied on a 70% confluent monolayer cultured in a well of a 12-well culture plate. After an incubation with siRNA-containing medium for 16 h, the cells were refreshed with regular culture medium, and incubated for a further 8 h. We plated these siRNA-transfected cells on a glass-bottomed 8-well chamber (ibidi) for the purpose of subsequent cell imaging, and grew the cells for ∼12 h. A second round of 8-h siRNA transfection (10 nM of si-RNA) was applied to these cells, and a subsequent 16-h incubation under regular culture medium. Cells were then fixed for further analysis.

Mouse monoclonal anti-E-cadherin antibody HECD-1 (ab1416, Abcam) was used to block the function of surface E-cadherin. For the fixed-cell imaging, the sparsely plated cell (2×10³ cells/cm² on glass bottom chamber slides) was cultured in the culture medium containing 5 μg/ml antibody HECD-1 for 1.5 h before fixation. For the live-cell imaging, the sparsely plated cell (2×10³ cells/cm²) was cultured in the HECD-1 containing culture medium (5 μg/ml) for 1 h prior to the time-lapse acquisition, followed with image acquisition for ∼2 h.

Actin de-polymerisor Cytochalasin D (Sigma-Aldrich) was used at 5 μM, and myosin II inhibitor Blebbistatin (Sigma-Aldrich) was used at 2.5 μM. For fixed imaging, the inhibitors were applied for 15 min under regular culture conditions before fixation.

MCF10A cells were grown on glass-bottomed 8-well chambers (ibidi), and fixed with 4% formaldehyde for 8 min at room temperature followed by permeabilisation with 0.1% Triton X-100 for 5 min. To ensure single cells, cells were seeded 1.5 days before fixation at a density of ∼2×10³ cells/cm². For confluent monolayers, cells were seeded at 1×10⁴ cells/cm² and grown for 2 days before fixation. The samples were then incubated with primary antibody overnight at 4°C as described in Table S1. Next, the samples were washed with PBS, and labelled with fluorochrome-conjugated secondary antibodies, DAPI (Thermo Fisher Scientific) and phalloidin (Abcam). The samples were examined using a FV3000 confocal microscope (Olympus) and 60× lens (1.30 NA, UPLSAPO, Olympus). To acquire images of the whole cell, scanning of multiple sections with a z distance of 0.5 μm was applied. These sections were either merged or converted into different orthogonal projections using Image J, and were also used in the analysis of spindle orientation. To analyse spindle orientation, the xz images were constructed using Image J, the tubulin image was used as a guide to determine the line crossing the two spindle poles to represent the orientation of the spindle, and the angle between this line and the horizontal was extracted. The images were processed with maximum intensity projection of three continuous slices for subcellular structures, such as retraction fibres and nascent junctions, using Image J. For the nascent junction measurements (width of the contact and intensity of different proteins of interest), the region of daughter–daughter contact was defined using F-actin staining.

For live-cell imaging, MCF10A cells were sparsely plated at ∼2×10³ cells/cm² on glass bottom chamber slides and cultured for 1.5 days. Before image acquisition, the cells were incubated in cell culture medium containing 250 nM SiR-DNA (Spirochrome) to label DNA for 30 min, which was then replaced with regular cell culture medium. The process of mitosis and cell division were imaged using an Olympus FV3000 confocal microscope (Olympus) and 20× lens (0.75 NA, UPLSAPO, Olympus) in a chamber maintained in 37°C and with 5% CO₂. Mitotic cells were identified by the rounded cell morphology and the bar shape of condensed chromosomes, and were recorded every 3 min for more than 30 min after anaphase onset. The z scanning, with 1 µm per z stage, was applied to cover the whole volume of the cell. The images were processed with average intensity projection for both bright field channel and the channel for SiR-DNA by Image J. To quantify the time taken by each daughter to re-adhere to the substrate, the time frame showing each daughter cell flattening was identified as the time daughter cells would ‘start spreading’), and the time of ‘Anaphase onset’ was subtracted as indicated in Fig. S2).

Confocal images were collected using an Olympus FV3000 confocal microscope and FV31S-SW Viewer software, and were processed using ImageJ (version 1.52p). Flow cytometry data were collected using a BD FACSAria III and software BD FACSDiVa (version 6.1.3), and were processed using FlowJo (version X). Python (version 3.8.1) was used for statistics. The following algorithms were used in Python for analysis and plotting: matplotlib (version 3.1.3), pandas (version 1.0.1), seaborn (version 0.10.0), numpy (version 1.18.1), and plotly (version 4.14.3).

The bar plots and the density plot are shown in a manner of median±quartiles. We used an unpaired two-tailed t-test to determine the significance between two distributions of variables using the built-in formula in Microsoft Office Excel (version 2016). The results were derived from at least three independent experiments.

This work was done on Wurundjeri land of the Kulin nation, and we pay our respects to the Elders past, present and emerging. We thank Michal Milgrom-Hoffman, Krystle Lim, and Rebecca Stephens for technical assistance and helpful discussions, and Mirren Charnley and John Lock for helpful comments on the manuscript.

The peer review history is available online at https://journals.biologists.com/jcs/lookup/doi/10.1242/jcs.260547.reviewer-comments.pdf