1. Cell Biology
  2. Developmental Biology
Download icon

Telophase correction refines division orientation in stratified epithelia

  1. Kendall J Lough
  2. Kevin M Byrd
  3. Carlos P Descovich
  4. Danielle C Spitzer
  5. Abby J Bergman
  6. Gerard MJ Beaudoin III
  7. Louis F Reichardt
  8. Scott E Williams  Is a corresponding author
  1. The University of North Carolina, United States
  2. The University of North Carolina School of Dentistry, United States
  3. University of California, San Francisco, United States
Research Article
  • Cited 3
  • Views 1,021
  • Annotations
Cite this article as: eLife 2019;8:e49249 doi: 10.7554/eLife.49249

Abstract

During organogenesis, precise control of spindle orientation balances proliferation and differentiation. In the developing murine epidermis, planar and perpendicular divisions yield symmetric and asymmetric fate outcomes, respectively. Classically, division axis specification involves centrosome migration and spindle rotation, events occurring early in mitosis. Here, we identify a novel orientation mechanism which corrects erroneous anaphase orientations during telophase. The directionality of reorientation correlates with the maintenance or loss of basal contact by the apical daughter. While the scaffolding protein LGN is known to determine initial spindle positioning, we show that LGN also functions during telophase to reorient oblique divisions toward perpendicular. The fidelity of telophase correction also relies on the tension-sensitive adherens junction proteins vinculin, α-E-catenin, and afadin. Failure of this corrective mechanism impacts tissue architecture, as persistent oblique divisions induce precocious, sustained differentiation. The division orientation plasticity provided by telophase correction may enable progenitors to adapt to local tissue needs.

Introduction

Stem and progenitor cells utilize asymmetric cell divisions to balance self-renewal and differentiation. Cell fate decisions can be influenced by the division axis, with the choice between symmetric and asymmetric fate outcomes dictated by positioning of the mitotic spindle. Mechanistically, precise control of division orientation may serve to equally or unequally partition fate determinants, or restrict access to a stem cell niche (Knoblich, 2008; Siller and Doe, 2009). Errors in division orientation can lead to defects in differentiation and cell identity, with the potential to drive overgrowths associated with cancer (Knoblich, 2010; Martin-Belmonte and Perez-Moreno, 2011; Neumüller and Knoblich, 2009).

The developing murine epidermis serves as an excellent model for studying how oriented cell divisions direct cell fate choices. Basal progenitors are capable of dividing either within the plane of the epithelium or perpendicular to it, resulting in symmetric or asymmetric divisions, respectively (Lechler and Fuchs, 2005; Smart, 1970). This process is governed by a conserved complex of spindle orienting proteins, including the essential linker LGN/Gpsm2 (Williams et al., 2011; Williams et al., 2014). During epidermal and oral epithelial stratification, LGN is recruited to the apical cortex in ~50% of mitoses, and LGN loss leads to increased planar divisions and severe differentiation defects (Byrd et al., 2016; Williams et al., 2011; Williams et al., 2014). Thus, a parsimonious explanation for the observed bimodal distribution of division angles is that perpendicular divisions occur when sufficient levels of LGN are recruited to the apical cortex during early mitosis, and planar divisions occur when this apical recruitment fails.

In this and other models, it is assumed that the division axis is established relatively early in mitosis, either through directed centrosome migration or spindle rotation. As an example of the former, in the Drosophila melanogaster testis and larval neuroblasts, one centrosome migrates to the opposite side of the cell during prophase, and the metaphase spindle forms along, and remains fixed by, this centrosomal axis (Rebollo et al., 2009; Siller et al., 2006; Yamashita et al., 2003). In other systems—including the C. elegans early embryo, D. melanogaster embryonic neuroblasts, and progenitors of the vertebrate neuroepithelia—the spindle dynamically rotates during metaphase to align with extrinsic niche-derived or intrinsic polarity cues (Geldmacher-Voss et al., 2003; Haydar et al., 2003; Hyman and White, 1987; Kaltschmidt et al., 2000). Collectively, these studies support the view that spindle orientation generally operates prior to anaphase onset.

On the other hand, there are hints from other studies that the metaphase-anaphase transition involves dynamic reorganization of the spindle orientation machinery. For example, in HeLa cells it has been shown that while LGN is essential for NuMA localization during early mitosis, LGN becomes dispensable during anaphase, when NuMA’s cortical localization is dependent upon phosphoinositides (Kotak et al., 2014). However, whether LGN functions to orient spindles at late stages of mitosis in other, polarized cell types, remains unknown.

Here, utilizing ex vivo live imaging in combination with mosaic RNAi, we find that division orientation in the developing murine epidermis is not determined solely by LGN localization during early mitosis. Surprisingly, LGN appears to play a "maintenance" role during anaphase/telophase, while an LGN-independent pathway involving adherens junction (AJ) proteins also acts to refine imprecise initial spindle positioning. We show that spindle orientation remains dynamic even into late stages of mitosis, and surprisingly, division axes remain random and uncommitted long after metaphase. While most cells enter anaphase with planar (0–30°) or perpendicular (60–90°) orientations and maintain this division axis through telophase, a significant proportion (30–40%) are initially oriented obliquely (30–60°), but undergo dramatic reorientation, a process we term telophase correction. In addition, we demonstrate that the α-E-catenin/vinculin/afadin cytoskeletal scaffolding complex is required for this correction to occur, and likely functions to modulate the tensile properties of the cell cortex by altering how actin is recruited to AJs. Mutants defective for telophase correction display precocious stratification which persists into later stages, highlighting the importance for this mechanism in generating normal tissue architecture. Furthermore, using genetic lineage tracing in afadin (Afdn) mutants, we confirm that uncorrected oblique divisions result in a strong bias toward differentiation over self-renewal.

Collectively, these studies support a novel two-step model of oriented cell division, where intrinsic factors such as LGN provide spatial cues that guide initial spindle positioning during early mitosis, while extrinsic factors such as cell-cell adhesions may provide a tension or density-sensing mechanism that refines the division plane during telophase to ensure normal tissue architecture. Our data further suggest that these mechanisms are modulated over developmental time to coordinate progenitor-expansive and differentiative programs.

Results

Randomized division orientation persists into anaphase

During peak stratification, epidermal basal cells undergo either LGN-dependent perpendicular divisions or LGN-independent planar divisions, with roughly equal frequency. LGN is invariably apical when recruited to the cell cortex during prophase and remains apical at telophase in perpendicular divisions (Williams et al., 2011; Williams et al., 2014). However, this bimodal distribution of division angles only emerges by ~E16.5, because the apical polarization of LGN is less efficient at earlier ages, resulting in a high proportion of oblique angles, and fewer perpendicular divisions (Williams et al., 2014). Our previous studies reported a bimodal distribution of division angles at late stages of mitosis and randomized division angles during metaphase (Williams et al., 2011; Williams et al., 2014), while other groups have reported that spindle rotation occurs during prometaphase and is fixed to a bimodal distribution by late metaphase/early anaphase (Poulson and Lechler, 2010; Seldin et al., 2016). While these studies agree that spindle rotation occurs, they come to different conclusions about when and how the spindle axis becomes fixed to a bimodal distribution.

Because these studies vary in the ages examined and the method used to identify mitotic cells at specific stages, we sought to apply a rigorous and unambiguous methodology to identify metaphase, anaphase and telophase cells at a single timepoint (E16.5), when nearly all divisions are either planar or perpendicular. Because phosphorylation at Ser10 and Ser28 of histone-H3 (pHH3) declines rapidly after metaphase (Hans and Dimitrov, 2001) antibodies raised against pHH3 vary in their ability to detect anaphase cells, and do not label telophase cells at all. Thus, we used another marker, Survivin (Birc5), which localizes to centromeres through metaphase, and redistributes to central spindle fibers and the cleavage furrow during anaphase and telophase, respectively (Beardmore et al., 2004; Caldas et al., 2005). In this manner, anaphase and telophase cells can be readily distinguished by their pattern of Survivin staining (Figure 1A). Since its original use (Williams et al., 2011), Survivin has been used by multiple groups across a variety of tissues to measure the division axis (Aragona et al., 2017; Asrani et al., 2017; Byrd et al., 2016; Cohen et al., 2019; Ding et al., 2016; Dor-On et al., 2017; Ellis et al., 2019; Ichijo et al., 2017; Jones et al., 2019; Liu et al., 2019; Niessen et al., 2013; Wang et al., 2018; Williams et al., 2014).

Figure 1 with 5 supplements see all
Telophase reorientation corrects oblique anaphase orientations.

(A) Sagittal sections from E16.5 embryos showing mitotic basal cells at indicated stages. Yellow arrows indicate division axis relative to basement membrane (dashed white line). Apical LGN (red) is generally present in oblique and perpendicular divisions, but absent from planar divisions. Survivin (green) is diffusely distributed between daughter pairs at anaphase, transitioning to stereotypic dual-puncta by telophase. (B) Radial histograms of division orientation at metaphase, anaphase, telophase and anaphase+telophase in E16.5 wild-type controls; n indicates number of divisions measured from >20 embryos per mitotic stage. (C) Same data as in (B), plotted as a cumulative frequency distribution. Note sigmoidal pattern at telophase (black, solid line), characteristic of bimodal distribution of division angles. Compare to linear pattern, characteristic of random distributions at metaphase (red) and anaphase (blue). (D) Schematic of experimental design for live imaging of embryonic epidermal explants. Krt14Cre; Rosa26mT/mG is used to label epidermis with membrane (m)-GFP and other tissues (including dermis) with mTdTomato. Alternatively, Krt14H2B-GFP is used to label nuclei while Rosa26mT/mG without Cre ubiquitously labels cells with membrane-tdTomato. (E) Z-projection stills from a movie of a Krt14Cre; Rosa26mT/mG (top) and Krt14H2B-GFP; Rosa26mT/mG (bottom) mitotic cell as it enters anaphase (defined as t = 0), through 60 min post-anaphase onset, depicting planar telophase correction. Epidermal-dermal boundary shown by red line. Dividing daughter pairs are outlined with yellow dashed lines. Division orientation angles are shown below (φ, anaphase onset; θ, +1 hr). (F) Traces of division orientation at five minute intervals for 15 representative cells from telophase onset to +1 hr. (G) Cumulative frequency distribution of division angles from Krt14Cre; Rosa26mT/mG live imaging experiments of E16.5 embryos at anpahse onset (blue; φ) and +1 hr later (black; θ). n indicates number of divisions from 4 embryos across four independent sessions. (H) Data from (G) depicting division orientations at anaphase onset and 1 hr later. Connecting lines demonstrate that ~60% oblique anaphase divisions reorient to planar (black lines) while the remaining ~40% correct to perpendicular (gray lines). Scale bars, 5 μm (A), 10 μm (E). **p<0.01, ***p<0.001, by Kolmogorov-Smirnov test. See also Figure 1—figure supplement 1.

We examined a large cohort of fixed sections of dorsal back skin epidermis from twenty-five E16.5 mouse embryos of varying strains (CD1, 129S4/SvJae and C57Bl6/J), and identified and imaged 536 Survivin+ metaphase, anaphase and telophase cells. We noted that anaphase and metaphase cells were comparatively rare, each occurring at ~1/5 the frequency of telophase cells (Figure 1B). In agreement with our previous observation, metaphase plates were oriented randomly, suggesting that spindle rotation occurs during metaphase. Surprisingly, however, the distribution of division angles remained random at anaphase, only establishing a bimodal distribution during telophase (Figure 1B,C). This trend held for each mouse strain (Figure 1—figure supplement 1A), demonstrating that differences in genetic background are unlikely to explain discrepancies in anaphase orientation reported by our group and others. Of note, because of the relative scarcity of anaphase divisions, they make a negligible contribution (Figure 1B, compare ‘telophase" to "ana+telo"), perhaps explaining why in previous studies, so few oblique divisions are reported in total Survivin+ pools. Nonetheless, these data demonstrate that basal cells remain uncommitted to a final plane of division beyond metaphase, and suggest that a previously uncharacterized spindle orientation mechanism occurs after anaphase onset.

Oblique anaphase divisions reorient during telophase

As a next step, we performed ex vivo live imaging of E16.5 embryonic epidermal explants (Cetera et al., 2018), in order to examine the dynamics of spindle orientation at late stages of mitosis. To easily discriminate the basal layer of the epidermis from the underlying dermis and visualize cell nuclei during mitosis we utilized two combinations of alleles: 1) Rosa26mT/mG + Krt14Cre, where cell membranes are GFP+ in the epidermis and tdTomato+ in the dermis, and 2) Rosa26mT/mG + Krt14H2B-GFP, where epidermal cell membranes are tdTomato+ and nuclei are H2B-GFP+ (Figure 1D). In both allele combinations, accurate measurements of the division angle relative to the epidermal-dermal border could be made in z-projections (Figure 1E). The Rosa26mT/mG + Krt14H2B-GFP combination was particularly useful for visualizing both the initiation of cleavage furrow ingression and the separation of nuclei that occurs at anaphase onset (defined as t = 0). Since cell nuclei could not be visualized in the Rosa26mT/mG + Krt14Cre background, we defined anaphase onset as the frame in which cleavage furrow ingression could be first visualized. Of note, in both allele combinations, the duration of anaphase was observed to be short—typically two 5’ frames elapsed where mGFP or mtdTom was not visible between daughter nuclei—providing an explanation for why anaphase cells were rarely observed in fixed tissue.

In both imaging paradigms, we observed a high proportion (~2/3) of basal cells which entered anaphase at either planar or perpendicular orientations that remained relatively fixed for the duration of the imaging period (Figure 1—figure supplement 1B, C; Figure 1—videos 1, 2). However, as suggested by our analyses of fixed tissue, many basal progenitors also frequently initiated anaphase at oblique angles (Figure 1E; t=0, φ=division angle). Notably, these oblique divisions invariably corrected to either planar or perpendicular within an hour (Figure 1E; Figure 1—figure supplement 1D, E; Figure 1—videos 3, 4). When the angle of division was plotted over time, we noted that this reorientation, hereafter referred to as telophase correction, generally occurred within the first 30 minutes of anaphase onset. (Figure 1F). Since little or no reorientation occurred after 1h, we assigned this as the imaging endpoint (t=+60min, θ=division angle). Of note, the distribution of division angles observed in these movies at anaphase onset (φ) and 1h later (θ) was remarkably similar to the distribution of anaphase and telophase orientations observed in fixed tissue (compare Figure 1G to 1C). When the behavior of individual cells was plotted at anaphase onset relative to 1h later, we observed that when φ>60°, correction tended to occur toward perpendicular, and when φ<30°, correction tended to occur toward planar, while oblique angles were less predictable (Figure 1H). This suggested that the directionality of correction is not purely stochastic.

LGN mediates perpendicular telophase correction

Previous studies have shown that LGN (Pins in Drosophila)—along with its binding partners Insc (Inscuteable), NuMA (Mud), and Gαi—play key roles in oriented cell divisions (Bowman et al., 2006; Du and Macara, 2004; Izumi et al., 2006; Kraut et al., 1996; Mora-Bermúdez et al., 2014; Schaefer et al., 2000; Siller et al., 2006; Williams et al., 2014; Zigman et al., 2005). In the conventional view, LGN functions primarily during prometaphase-metaphase by facilitating capture and anchoring of astral microtubules to the cell cortex. In developing stratified epithelia, LGN first localizes to the apical cortex during prophase (Byrd et al., 2016; Lechler and Fuchs, 2005; Williams et al., 2011; Williams et al., 2014). However, our finding that a large proportion of anaphase cells are oriented obliquely suggests that initial perpendicular spindle positioning by LGN may be imprecise, and raises the question of whether LGN may also function during perpendicular telophase correction.

To test this, we performed ex vivo live imaging of Krt14Cre; Rosa26mT/mG epidermal explants mosaically-transduced with a previously validated shRNA targeting LGN/Gpsm2 (Gpsm21617) or non-targeting Scramble shRNA control (Williams et al., 2011; Williams et al., 2014). The H2B-RFP reporter allowed us to track pronuclear separation during anaphase onset, and distinguish RFP+ transduced/knockdown basal cells from non-transduced/wild-type RFP- internal controls (Figure 2A,B). Like wild-type explants, Scramble RFP+ and Gpsm21617 RFP- control cells were randomly oriented at anaphase onset, but corrected to a bimodal distribution 1h later (Figure 2C–E). Compared to wild-type cells, a higher proportion of Gpsm21617 RFP+ cells entered anaphase at planar (φ <30) orientations (75% vs 30% for Gpsm21617 RFP- and 32% for Scramble RFP+). In addition, very few Gpsm21617 RFP+ cells (2%, n = 49) entered anaphase at perpendicular (φ>60) orientations (Figure 2D,E). These data support our previous findings that LGN is required for initial positioning of perpendicular spindles. Interestingly, however, the minority (23%) of Gpsm21617 RFP+ cells that entered anaphase at oblique angles invariably corrected toward planar (Figure 2D,E). Taken together, these data suggest that, in addition to its known role in orienting spindles along the apicobasal axis during prometaphase, LGN also serves a second maintenance function later in mitosis, where it promotes perpendicular correction during telophase.

LGN mediates perpendicular but not planar telophase correction.

(A) Schematic of modified experimental protocol of live imaging of epidermal explants (see Figure 1D) incorporating lentiviral shRNA transduction to generate mosaic knockdown tissue. Transduced/knockdown regions are marked with histone H2B-mRFP1 (H2B-RFP). (B) Stills from live imaging of Scramble (top) or Gpsm21617 H2B-RFP+ cells (bottom) undergoing planar correction, annotated as in Figure 1E. (C,D) Cumulative frequency distributions of division orientation from (C) Scramble or (D) Gpsm21617 H2B-RFP (+/-) live imaging experiments at anaphase onset (φ) and one hour later (θ). Scramble RFP+ and Gpsm21617 RFP- cells display similar patterns of telophase correction as observed in wild-type explants (Figure 1G). While Gpsm21617 RFP+ cells are more biased toward planar/oblique at anaphase onset, significant planar correction still occurs; n indicates observed divisions from five embryos imaged in four technical replicates. (E) Data from (C,D) depicting orientation at anaphase onset (φ) and 1 hr later (θ) for Scramble RFP+ and Gpsm21617 RFP- and RFP+ cells. ~ 95% of LGN knockdown cells correct to planar (<30°) 1 hr later. Scale bars, 10 μm. *p<0.05 by Kolmogorov-Smirnov test.

Directionality of telophase correction is correlated with basement membrane contact

We next sought to address the mechanisms underlying planar directed telophase correction. In our wild-type live imaging experiments, we observed that while initial orientations of φ >60° typically corrected to perpendicular, and φ <30° to planar, the behavior of intermediate orientations (φ = 30–60°) was less predictable (Figure 1H). However, we noted that apical daughters undergoing planar telophase correction frequently displayed a unique, balloon-shaped morphology and appeared to maintain contact with the basement membrane (open arrowheads in Figure 3A). Remarkably, maintenance of this basal endfoot predicted planar reorientation, while loss of contact predicted the opposite (Figure 3B,C; Figure 3—figure supplement 1A). Importantly, this correlation between basal contact and telophase correction was unaltered by expression of Scramble or Gpsm21617 shRNAs (Figure 3—figure supplement 1B,C). These data suggest that transient oblique metaphase-anaphase orientations are corrected in a manner dependent on whether they retain contact with the basement membrane following cleavage furrow ingression.

Figure 3 with 1 supplement see all
Maintenance of basal contact correlates with planar-directed telophase correction.

(A) (top) z-projection stills from a movie of a mitotic cell as it enters anaphase (t = 0) through 60 min post-anaphase onset, depicting planar telophase correction. Epidermal-dermal boundary shown by red line. Dividing daughter pairs are outlined with yellow dashed lines. Division orientation angles are shown below (φ, telophase onset; θ, +1h). (bottom), xz en face views at same timepoints. Yellow and white arrowheads indicate plane of optical section for apical and basal daughters, respectively. In most cases, planar correction is preceded by maintenance of basement membrane contact (open arrowheads), which are most apparent in the en face basal focal plane, where they appear as small membrane circles. (B) Data from Figure 1G,H sorted based on presence or absence of basal contact. Connecting lines demonstrate that oblique-dividing daughters retaining basal contact correct towards a planar orientation, while those losing contact correct towards perpendicular. (C) Data from (B) demonstrating that the degree of correction correlates with initial anaphase orientation. (D) Whole mount imaging of wild-type E16.5 epidermis stained with phalloidin and phosphorylated myosin-light chain 2 (pMLC2). Orthogonal views (top) of DAPI highlight oblique division orientation. The basal endfoot observed in live imaging of telophase correction (see panel A) can be observed in the basal en face view. Pair-wise measurements (inset graph) of pMLC2 at the cell cortex in the apical plane and basal endfoot of oblique divisions are connected by the gray line. (E) Cartoon representation of tension-sensitive model of AJ assembly. In the absence of tension, α-E-catenin exists in an autoinhibited closed conformation, masking the α18 epitope. In the presence of actin-mediated tension, α-E-catenin opens, exposing the α18 epitope and vinculin binding domain. (F) Whole mount images prepared as in (D) stained with total α-E-catenin and open conformation-specific α18 antibody. Pair-wise measurements (inset graph) of α18 at the cell cortex in the apical plane and basal endfoot of oblique daughter cells are connected by the gray line demonstrates increased open or ‘tensile’ α-E-catenin in the basal endfoot. (G) Quantification of α18: α-E-catenin fluorescence intensity ratio in variable division types or stages of mitosis. Anisotropy is greatest in oblique divisions between the basal endfoot and apical cortex of the oblique daughter cell. Scale bars, 10 μm (A,D,F). P values determined by Wilcoxon test (D,F). *p<0.05, **p<0.01. See also Figure 3—figure supplement 1.

Telophase corrective basal contacts display hallmarks of elevated actomyosin contractility

Given the dynamic changes to cell shape that occur during telophase correction, we hypothesized that they may correlate with distinct molecular changes in the underlying actomyosin cytoskeleton. To test this, we performed immunostaining on E16.5 epidermal whole mounts for actin (phalloidin) and active phosphorylated (Ser19) myosin light chain II (pMLC2), and identified rare, oblique divisions with the characteristic basal endfoot. Interestingly, the intensity of pMLC2 was higher specifically in the endfoot process compared to the apical cortex of the same daughter cell, while actin levels were similar (Figure 3D; Figure 3—figure supplement 1D). This anisotropy suggests that the basal endfoot may be enriched in contractile actomyosin, which we speculate may serve the function of pulling the apical daughter back into the basal layer.

Increased actomyosin contractility can be indicative of elevated tension across adherens junctions (AJs) that anchor the cytoskeleton to the cell membrane. The AJ is canonically composed of transmembrane cadherins, which couple neighboring cells through trans-dimerization in the extracellular space and link to the underlying actin-cytoskeleton via α-E-catenin (Ratheesh and Yap, 2012). In the presence of actin-dependent tension, α-E-catenin undergoes a conformational change, exposing an epitope within its mechanosensitive modulatory (M) domain that is recognized by the α18 antibody (Buckley et al., 2014; Hansen et al., 2013; Rübsam et al., 2017; Yonemura et al., 2010) (Figure 3E). To investigate whether α-E-catenin undergoes conformational changes during telophase correction, we performed whole mount immunofluorescence for total and tensile α-E-catenin, seeking out rare anaphase cells undergoing oblique divisions. In agreement with the observed increase in pMLC2 in the basal endfoot of oblique telophase cells, levels of α18 were also higher in the basal endfoot compared to the apical cell cortex in these cells (Figure 3F). Importantly, this increased intensity was specific to the α18 epitope as levels of total α-E-catenin did not display similar anisotropy (Figure 3—figure supplement 1E). This elevated α18:α-E-catenin ratio was only observed in the basal endfoot of oblique divisions, and not in planar divisions, metaphase cells, or non-mitotic neighbors (Figure 3G). These data suggest that increased actomyosin contractility and associated conformational changes to α-E-catenin could play a role in planar directed telophase correction.

The actin-binding protein, vinculin, regulates α-E-catenin conformation

α-E-catenin (Ctnna1) serves as the core mechanosensor at AJs, such that force across AJs induces a conformational change in α-E-catenin which exposes a vinculin-binding domain within the M region (Ladoux et al., 2015; Yonemura et al., 2010). The binding of α-E-catenin to both actin and vinculin (Vcl)—another cytoplasmic actin-binding protein that functions at both AJs and focal adhesions—is force dependent, and vinculin and α-E-catenin cooperate to strengthen AJ-mediated adhesion (Choi et al., 2012; Huang et al., 2017; Seddiki et al., 2018; Thomas et al., 2013; Weiss et al., 1998; Yao et al., 2014). Other studies have shown that the actin scaffold afadin (Afdn) is capable of binding directly to α-E-catenin via an internal domain proximal to the vinculin binding domain, and that afadin is recruited to sites of α-E-catenin activation together with vinculin (Mandai et al., 1997; Matsuzawa et al., 2018; Pokutta et al., 2002).

Due to the challenges of finding rare oblique-correcting cells in vivo, to further investigate the interplay between AJ complex proteins and actomyosin contractility, we turned to a calcium-shift adhesion assay in primary cultured keratinocytes (Vasioukhin et al., 2000). Following 8h of exposure to 1.5 mM (high) Ca2+, Scramble control keratinocytes form linear AJs containing both vinculin and α-E-catenin (Figure 4A). Ctnna1 knockdown led to a reduction in junctional vinculin, while Vcl knockdown led to a reduced fluorescence intensity ratio of α18 ("tensile") to total α-E-catenin (Figure 4A–C), confirming that the tension sensitivity of α-E-catenin is vinculin-dependent in keratinocytes. Interestingly, while Vcl loss reduced the proportion of tensile α-E-catenin, this was a result of a net increase in total α-E-catenin, while total α18 intensity remained unchanged or even increased (Figure 4—figure supplement 1A). This suggests that higher levels of junctional α-E-catenin may partially compensate for Vcl loss to maintain a threshold level of tensile α-E-catenin.

Figure 4 with 1 supplement see all
Vinculin and afadin regulate α-E-catenin conformation and AJ linkage to the actin cytoskeleton.

(A) Stable primary murine keratinocytes cell lines grown in the presence of high (1.5 mM) Ca2+ for 8h form nascent cell-cell adhesions, stained for total α-E-catenin (green); open, "tensile" α-E-catenin (α18, red); and vinculin (gray). Single junction magnifications (yellow dashed region) shown below, demonstrate that Vcl knockdown results in a reduced α18: α-E-catenin ratio, quantified in (B,C). (B) Fluorescence intensity quantification of junctional vinculin in Scramble and Ctnna1 knockdown keratinocytes. Loss of Ctnna1 reduces vinculin accumulation in nascent AJs. (C) Quantification of α18:α-E-catenin fluorescence intensity ratio in Scramble and two independent Vcl shRNA cell lines. Vcl knockdown reduces the proportion of α-E-catenin in the open conformation. (D) Primary mouse keratinocytes after 8h Ca2+ shift—labeled with phalloidin (red) and E-cad (green)—which accumulate in linear bands at cell-cell junctions in Scramble control cells. Yellow boxed region shown at high magnification below; n indicates junctions evaluated. Vcl and Ctnna1 knockdown cells show defects in linear actin accumulation and immature "zipper" junctional morphology. (E) Junction continuity quantification based on % of junction length above threshold for E-cad (see Materials and methods). Loss of Vcl or Ctnna1 reduces junction continuity. (F) Afdnfl/fl primary keratinocytes mosaically infected with Cre-RFP (red) after 8h 1.5 mM Ca2+ shift, stained for E-cad (green), afadin (red), and phalloidin (gray). Junctions between two uninfected cells (WT:WT) show linear morphology with consistent E-cad (green), afadin (red) and phalloidin (gray) labeling. In contrast, junctions between two infected cells are punctate, with less junction-associated phalloidin. (G) Quantification of E-cad continuity along junction length, as in (E). (H) Quantification of fluorescence intensity of actin (phalloidin) measured by orthogonal linescans. Phalloidin is decentralized in KO:KO junctions (red) compared to WT:WT (black;. n indicates junctions evaluated. (I) Quantification of α18:α-E-catenin fluorescence intensity ratios from homogenous Afdnfl/fl, Afdn-cKO, Scramble, and Afdn2711 primary keratinocytes stained as in (A); n indicates junctions analyzed. Scale bars, 20 μm or 5 μm (junctional insets). P values determined by student’s unpaired t-test; ***p<0.001. See also Figure 4—figure supplement 1.

While α-E-catenin is still recruited to AJs in Vcl knockdown keratinocytes, Vcl-deficient junctions appeared abnormal, in agreement with a recent report (Rübsam et al., 2017). Vcl-deficient junctions were wider and more punctate than controls, with a morphology reminiscent of immature "spot AJs" or "adhesion zippers" (Vasioukhin et al., 2000). In wild-type keratinocytes cultured for 30 min in high Ca2+, nascent cell-cell junctions displayed discontinuous E-cadherin puncta associated with loosely-organized radial actin filaments, while after 8h, E-cadherin and actin became tightly associated in a circumferential belt (Figure 4—figure supplement 1B). We developed a quantitative method to measure E-cadherin puncta as a means of assessing junctional maturation, such that higher "continuity" values represent mature linear junctions (e.g., 8h Ca2+ shift) while lower continuity values represent spot junctions (e.g., 30 min Ca2+ shift) (Figure 4—figure supplement 1C). Similar to wild-type controls, following an 8h Ca2+ shift, Scramble keratinocytes displayed junctions with linear actin that was closely aligned with E-cadherin. In contrast, in Ctnna1 and Vcl knockdown keratinocytes E-cadherin was punctate and discontinuous, and displaced from the cortical actin belts (Figure 4D,E). These data demonstrate that α-E-catenin and vinculin are required for the proper maturation of AJs.

Afadin is required for normal AJ morphology and is a novel regulator of α-E-catenin conformation

Afadin and its Drosophila homolog Canoe (Cno) are required to stabilize actin-AJ associations during moments of high actomyosin contractility, suggesting a role in establishing/maintaining tensile loads (Choi et al., 2016; Sawyer et al., 2011). To examine whether afadin loss influences AJ-associated actin in keratinocytes, we generated mosaic cultures of wild-type and Afdnfl/fl cells transduced with lentiviral Cre-RFP (Figure 4F). E-cadherin+ AJs between two wild-type uninfected cells (WT:WT) showed normal junctional accumulation of afadin, while AJs between two RFP+ cells (KO:KO) lacked afadin (Figure 4F, red). Afdnfl/fl Cre-RFP+ cells also demonstrated increased levels of cytoplasmic E-cadherin, and KO:KO junctions displayed punctate, rather than linear, E-cadherin (Figure 4G), reminiscent of immature "spot" junctions. In addition, while WT:WT junctions showed tight association of actin with E-cadherin, like Vcl and Ctnna1912 knockdown AJs, Afdn KO:KO AJs showed reduced junctional actin, with actin bundles frequently displaced ~1 μm from the junction (Figure 4F,H). These data suggest that afadin plays an essential role in linking cortical actin to the AJ complex, with potential consequences on E-cadherin clustering.

Since it has been shown that AJ components such as E-cadherin regulate junctional recruitment of vinculin from focal adhesions in a tension dependent manner (Noethel et al., 2018; Rübsam et al., 2017), and we noted that α-E-catenin and vinculin are required for afadin accumulation in the AJ (Figure 4—figure supplement 1D,E), we wondered whether afadin reciprocally regulates α-E-catenin or vinculin. Similar to observations in Vcl knockdown, knockout or knockdown of Afdn resulted in increased junctional accumulation of α-E-catenin, with no observable increase in the α18 epitope, reducing the α18:α-E-catenin fluorescence intensity ratios (Figure 4I; Figure 4—figure supplement 1F–H). Importantly, loss of Afdn also reduced vinculin accumulation in the junction, highlighting a reciprocal regulatory relationship (Figure 4—figure supplement 1I,J). Collectively, these data suggest that afadin is a novel regulator of AJ maturation by affecting α-E-catenin conformation and vinculin recruitment.

Ctnna1, Vcl and Afdn knockdown leads to randomized division orientation

The enrichment of pMLC2 and tensile α-E-catenin in the basal endfoot that we observed in vivo—in addition to the aberrant adhesion and actin organization that we observed in vitro in Ctnna1, Vcl and Afdn mutants—prompted us to investigate whether loss of AJ components alters spindle orientation. To this end, we utilized Survivin to label late-stage mitotic cells and integrin-β4 to label the basement membrane to assess division orientation in E16.5 fixed back skin sections where AJ components where knocked down using our in utero lentiviral delivery method (Ctnna1, Vcl and Afdn), or conditionally knocked out in the epidermis (Afdn) (Figure 5—figure supplement 1A). We first confirmed the efficacy of knockdown/knockout in vivo using antibodies specific to α-E-catenin, vinculin, and afadin (Figure 5A-C). Each AJ protein was localized to the lateral and apical cortex in WT basal cells, as well as to cell membranes in differentiated suprabasal cells. This staining was strongly reduced in RFP+ regions transduced with each shRNA and eliminated in regions where Afdn was knocked out by either lentiviral-mediated delivery of Cre-RFP or by conditional deletion using Krt14Cre; Afdnfl/fl (hereafter referred to as Afdn cKO) (Figure 5A-C).

Figure 5 with 1 supplement see all
The α-E-catenin/vinculin/afadin pathway is required for normal division orientation.

(A) Immunofluorescent images taken from E16.5 sagittal sections of wild-type littermate controls (left) or embryos transduced with Ctnna1912 H2B-RFP (right). Epidermal junctional α-E-catenin (green) is lost in Ctnna1912 RFP+ epidermis. (B) E16.5 epidermis infected with Vcl3466 H2B-RFP (red) and stained rabbit with anti-vinculin antibody. While suprabasal staining is dramatically reduced in infected samples, some non-specific cytoplasmic basal-layer staining remains. (C) Afadin (green) and E-cadherin (red) immunostaining in E16.5 sections. Mosaic region of Afdn2711 H2B-RFP (top panel) or Cre-RFP (in Afdnfl/fl embryo; bottom panel) lentiviral transduction. Region of high transduction (red line) demonstrates efficient loss of junctional afadin signal, spared in region of low transduction (white line). E16.5 Afdnfl/fl controls (right, top) with conditional deletion mediated by Krt14Cre (cKO) (right, bottom). (D–I) Cumulative frequency distributions of telophase division angles from fixed E16.5 sections of shRNA knockdown samples and littermate controls. (D) Ctnna1912 knockdown (red) and control littermates (black); n indicates measurements from 6 to 7 independent embryos. (E) Vcl2803 H2B-RFP mosaic samples showing RFP+ mutants (red) alongside RFP- internal (gray) and wild-type littermate (black) controls; n indicates measurements from 3 to 4 independent embryos. (F) Vcl3466 H2B-RFP mosaic samples shown as in E; n indicates measurements from 3 to 4 independent embryos. (G) Afdn2711 H2B-RFP mosaic samples shown as in E-F; n indicates measurements from 3 to 6 independent embryos. (H) Afdnfl/fl Cre-RFP samples (red) shown alongside uninjected littermates (black); n indicates measurements from 3 to 4 independent embryos. (I) Cumulative frequency distribution of E16.5 telophase division angles in Afdnfl/fl, Afdn cKO, and Afdn cKO + Vcl3466 H2B-RFP epidermis. Vinculin knockdown does not exacerbate Afdn knockout phenotype. Scale bars, 20 μm (A–C). P values determined by Kolmogorov-Smirnov test (D–I). *p<0.05, **p<0.01. See also Figure 5—figure supplement 1.

In each AJ knockdown cohort, we observed a normal bimodal distribution of division angles in wild-type littermate and non-transduced RFP- controls in late stage mitotic cells. However, RFP+ cells displayed randomized division orientation (Figure 5D–G), similar to what we observed at anaphase onset in fixed tissue and live imaging. We further validated this phenotype using Afdnfl/fl embryos (Beaudoin et al., 2012), and confirmed that division orientation was randomized whether Afdn was deleted by lentiviral delivery of Cre-RFP or transgenic expression of Krt14Cre, and analyzed in either sections or wholemounts (Figure 5H; Figure 5—figure supplement 1B,C). Finally, because both afadin and vinculin interact directly with α-E-catenin, we sought to test genetically whether afadin and vinculin operate in the same molecular pathway. To do so, we performed embryonic lentiviral injection of the Vcl3466 shRNA on an Afdn cKO or Afdnfl/fl background. Examination of division orientation in single and double mutants revealed that vinculin loss did not exacerbate the Afdn cKO phenotype, suggesting that these proteins do not act additively in the context of division orientation (Figure 5I).

Tension-sensitive components of the AJ are essential for telophase correction

We next sought to address whether the randomized division orientation phenotype observed in AJ mutants was due to errors in initial spindle positioning or telophase correction. To this end, we performed live imaging of lentiviral-transduced Ctnna1912, Vcl3466 and Afdn2711 H2B-mRFP1 epidermal explants on a Krt14Cre; Rosa26mT/mG background. We began with α-E-catenin, because it had previously been shown that Ctnna1 loss leads to randomized division orientation in the developing epidermis (Lechler and Fuchs, 2005). As observed earlier with wild-type and Gpsm21617 cells, many Ctnna1912 RFP+ basal cells entered anaphase at oblique orientations, with the apical daughter possessing a basal endfoot extending to the basement membrane (Figure 6A; Figure 6—figure supplement 1A; Figure 6—video 1). Because of the high efficiency of transduction achieved with the Ctnna1 lentivirus in these experiments, we utilized wild-type littermates as controls rather than RFP- cells, which were rare. We imaged 74 Ctnna1 RFP+ mitotic cells, and observed that α-E-catenin loss had no effect on initial anaphase orientation, which was randomized, akin to wild-type littermates. However, while wild-type cells corrected to a bimodal distribution within 1h of anaphase onset, there was no change in the distribution of division angles in Ctnna1 RFP+ cells between anaphase onset and 1h later (Figure 6B). Whether or not apical daughters maintained basal contact, there appeared to be no obvious pattern to the directionality of telophase reorientation, with a majority of cells showing little or no change over 1h following anaphase onset (Figure 6C; Figure 6—figure supplement 1B,F).

Figure 6 with 4 supplements see all
AJ mutants fail at both planar and perpendicular telophase correction.

(A) Movie stills of Ctnna1912 RFP+ mitotic cell, annotated as in Figure 3A. While the presence of basal contact (open arrowhead) would predict planar correction, this division remains oblique when reevaluated 1h later. (B) Cumulative frequency distribution of division angles from E16.5 live imaging experiments of Ctnna1912 RFP+ and wild-type littermates; n values indicate cells imaged from three embryos images in two separate sessions. (C) Division orientation at anaphase onset (φ) and 1h later (θ) for Ctnna1912 knockdown and wild-type littermates, plotted from data in (B). Ctnna1912 RFP+ cells show no obvious correction pattern. (D) Cumulative frequency distribution of division orientation at anaphase onset (φ) and 1hrpost-anaphase (θ) for RFP+ and RFP- populations, from movies of Vcl3466 mosaic tissue; n indicates divisions from four embryos imaged in three separate sessions. (E) Data from (D) depicting orientation at anaphase onset (φ) and 1h later (θ) for RFP- and RFP+ cells. RFP- controls sort anaphase orientation (φ) into bimodal distribution within 1h(θ) in a basal-contact dependent manner; Vcl3466 RFP+ cells display minimal change, or correct irrespective of basal contact. (F) An obliquely-oriented Afdn2711 RFP+ cell fails to reorient, while losing basal contact (open arrowhead). (G) Timelines of division orientation at 5 min intervals from movies of Afdn2711 RFP- (black) and RFP+ (red) for 15 representative cells per group. Telophase reorientation establishes bimodal distribution within ~30 min in RFP- control cells that enter anaphase at oblique angles, while RFP+ cells fail to demonstrate any sorting behavior over a full hour following anaphase onset. (H) Cumulative frequency distributions of division orientation from E16.5 live imaging of Afdn2711 RFP+ and wild-type littermates; n indicates observed divisions from three embryos imaged in two separate sessions. (I) Radial change (φ-θ) for oblique anaphase divisions (30°−60°) in several shRNA conditions. While loss of LGN allows for normal telophase correction, Afdn, Ctnna1, and Vcl knockdown results in incoherent or minimal radial change; n indicates number of divisions from 3 to 6 individuals embryos images in 2–4 technical replicates. (J) Division orientation at anaphase onset (φ) and one hour later (θ) for Afdn2711 RFP+ and RFP- cells, plotted from data in (C). RFP- controls correct into a bimodal distribution, while RFP+ cells reorient randomly. Scale bars, 10 μm. P values determined by Kolmogorov-Smirnov test (B,D,H) or student’s t-test (I). *p<0.05, **p<0.01, ***p<0.001. See also Figure 6—figure supplement 1.

Like Ctnna1912 knockdown cells, Vcl3466 RFP+ cells frequently entered anaphase at oblique orientations and showed little movement during telophase (Figure 6—figure supplement 1C; Figure 6—video 2). RFP- cells corrected to a bimodal distribution, although in these experiments a higher proportion of planar corrections were observed than in previous studies, perhaps due to the slight differences in their developmental stage (Figure 6D). Nevertheless, as a population, Vcl3466 RFP+ cells displayed a randomized distribution of division angles at both anaphase onset and 1h later (Figure 6D). As with Ctnna1 loss, Vcl knockdown reduced the magnitude of telophase reorientation, and eliminated the predictiveness of basal contacts for correction directionality, causing failure in both perpendicular and planar correction (Figure 6E; Figure 6—figure supplement 1D,F).

In Drosophila, the afadin homologue Cno is essential for asymmetric cell division of embryonic neuroblasts (Speicher et al., 2008). Recent studies in mammals have similarly described a role for afadin in regulating division orientation in the embryonic kidney and cerebral cortex (Gao et al., 2017; Rakotomamonjy et al., 2017). In fixed tissue, we knocked down or knocked out Afdn by three different methods, each resulting in randomized division orientation in Survivin+ late-stage mitotic cells (Figure 5G,H; Figure 5—figure supplement 1). Because the native fluorescence of Cre-RFP is dim and photobleaches rapidly, and live imaging of Afdn knockouts requires a complex breeding scheme involving four alleles, we utilized the Afdn2711 shRNA for ex vivo imaging experiments (Figure 6F). As with α-E-catenin and vinculin loss, Afdn knockdown had no effect on initial anaphase orientation, while oblique divisions failed to undergo either planar or perpendicular-directed telophase correction (Figure 6G,H; Figure 6—figure supplement 1E,F; Figure 6—video 3). Afdn knockdown phenocopied loss of vinculin and α-E-catenin, with minimal or randomized reorientation of oblique divisions (Figure 6I). Moreover, while endfoot contact at anaphase onset was predictive of telophase correction directionality in RFP- cells, this was not the case in Afdn2711 RFP+ cells (Figure 6J). Notably, however, while oblique Ctnna1912 and Vcl3466 cells generally retained basal endfoot processes if they were present at anaphase onset, 73% of oblique Afdn2711 cells lost contact during telophase, suggesting that afadin may function in endfoot retention. Collectively, these studies demonstrate that mechanosensitive AJ proteins do not appear to function in initial spindle positioning, but play important modulatory roles in mediating telophase correction, which, when disrupted, lead to persistent division orientation errors.

Telophase correction occurs independently of canonical polarity and spindle-orienting cues

The Drosophila afadin ortholog Cno is essential for early establishment of apical-basal polarity during cellularization (Bonello et al., 2018; Choi et al., 2013). A similar role has been described for afadin in mammalian development (Komura et al., 2008; Rakotomamonjy et al., 2017; Yang et al., 2013). Furthermore, both Par3 and its Drosophila ortholog Bazooka are required for oriented cell divisions via regulation of LGN localization (Schober et al., 1999; Williams et al., 2014; Wodarz et al., 1999). Thus, we asked whether Afdn loss impacts expression of the canonical apical polarity cue Par3. In Afdnfl/fl controls, Par3 accumulates at the apical cortex throughout the cell cycle (Figure 7—figure supplement 1A). We measured Par3 radial fluorescence intensity at interphase and determined that Afdn cKO epidermis shows a 15–30% reduction in apical accumulation (Figure 7—figure supplement 1A,B). However, this had no effect on the apical positioning of centrosomes (Figure 7—figure supplement 1C,D), suggesting that apical-basal polarity remains largely intact in Afdn mutants.

Previous studies have shown that Cno interacts directly with the LGN ortholog Pins and regulates its cortical recruitment (Speicher et al., 2008; Wee et al., 2011). Mammalian afadin and LGN also directly interact in HeLa cells, where they function to promote planar divisions (Carminati et al., 2016). In addition, E-cadherin is capable of regulating division orientation through a direct interaction with LGN (Gloerich et al., 2017; Hart et al., 2017). Finally, Ctnna1 knockout has been reported to perturb LGN localization in epidermal basal cells (Lechler and Fuchs, 2005). These studies suggested that the division orientation defects we observed in Ctnna1 and Afdn mutants could be due to mislocalized LGN.

Using pHH3 to label cells in early mitosis in Afdnfl/fl controls and Afdn cKO mutants, we observed similar patterns of LGN crescent localization, cortical intensity, and efficiency of apical polarization (Figure 7A–D). In addition, we did not observe any obvious or significant changes to LGN localization in Afdn2711, Ctnna1912 or Vcl3466 cells (Figure 7C,D; Figure 7—figure supplement 1E). Thus, AJ components appear to be dispensable for initial apical positioning of LGN. In Drosophila neuroblasts, genetic epistasis and protein localization studies support the view that Cno/afadin acts downstream of Pins/LGN and upstream of Mud/NuMA (Speicher et al., 2008). However, we find that neither NuMA, nor its downstream binding partner dynactin, appears to be mislocalized in Afdn mutants (Figure 7—figure supplement 1F–H). In addition, NuMA staining overlapped with LGN in early mitotic cells, regardless of afadin presence/absence (91% in Afdnfl/fl, n = 22; 93% in Afdn cKO, n = 14). These data suggest that afadin plays little, if any role, in regulating the LGN-NuMA-dynactin pathway during initial spindle positioning.

Figure 7 with 2 supplements see all
AJ mutants alter division orientation via LGN-independent mechanisms.

(A) Immunostaining for LGN (green) in E16.5 Afdnfl/fl and Afdn-cKO epidermis. LGN localizes at the apical cortex during mitosis regardless of afadin loss. (B) Quantification of LGN radial fluorescence intensity in E16.5 mitotic cells; n indicates LGN+ mitoses from 2 to 3 independent embryos. (C) Orientation of LGN crescents in E16.5 mitotic cells from indicated groups. Knockdown or knockout of AJ components does not significantly alter the tendency of LGN to localize apically. (D) (top) LGN (red) localization patterns in mitotic (green) basal keratinocytes. (Bottom) Quantification of LGN rate of recruitment, binned by genotype. LGN localizes to the apical cortex in ~50% of mitoses (black/red), is absent in ~45% (gray/pink), and "other" in the remaining ~5% (white), remaining unchanged in AJ knockdown/knockout mutants; n indicates mitotic cells from 2 to 3 independent embryos. (E) Costaining of E16.5 metaphase (left) and telophase (right) divisions with α-tubulin (red) and LGN (green) in Afdn-cKO (bottom) and Afdnfl/fl control littermates (top). (F) Quantification of the deviation between the metaphase spindle or division axis (red arrow in E) and LGN radial orientation (green arrow in E). Afdn knockout does not disrupt early spindle-LGN linkage, but shows oblique telophase orientation despite normal localization of LGN. (G) Immunostaining for the differentiation marker K10 (green) and lentiviral H2B-RFP reporter simultaneously with β4-integrin (red) in E16.5 Gpsm21617 infected embryos with an Afdnfl/fl (left) or Afdn-cKO (right) background. Dual loss of Afdn and Gpsm2 results in increased stratification relative to Gpsm2 loss alone. (H) Quantification of spinous layer (K10+) thickness from images as in (G). (I) Cumulative frequency distribution of telophase division angles from fixed sagittal sections of E16.5 embryos. n indicates number of divisions from 2 to 3 independent embryos. Scale bars, 5 μm (A,D,E), 25 μm (G). **p<0.01, ***p<0.001, determined by Kolmogorov-Smirnov test (I) or student’s t-test (F,H). See also Figure 7—figure supplements 1 and 2.

We previously demonstrated that the mitotic spindle can become misaligned with cortical LGN during metaphase, for example following NuMA (Numa1) knockdown (Williams et al., 2011). Thus, we sought to examine whether Afdn loss could also lead to uncoupling of the division axis from LGN polarity cues during mitosis, perhaps independently of NuMA. To test this, we co-stained prefixed E16.5 Afdnfl/fl and Afdn cKO sections with LGN and α-tubulin in order to visualize spindles during metaphase, and cleavage furrow ingression later in telophase (Figure 7E). Importantly, while afadin loss altered telophase division orientation, it had no effect during metaphase, where spindles were randomly-oriented (Figure 7—figure supplement 2A). Furthermore, the apical LGN crescent aligned with the metaphase spindle axis—regardless of its orientation—in both Afdn cKO and control embryos (Figure 7E,F; Figure 7—figure supplement 2B). By all these metrics, LGN localization was also unperturbed in Afdn2711, Ctnna1912, Vcl2803 and Vcl3466 knockdowns as well (Figure 7C–F). However, in telophase cells, while LGN remained apically-positioned in both controls and Afdn mutants, the orientation of the spindle axis became uncoupled from LGN in Afdn mutants (Figure 7F; Figure 7—figure supplement 2B,D). Similarly, knockdown of α-E-catenin or vinculin phenocopied afadin loss, demonstrating that AJ perturbation does not alter LGN localization, but does affect the ability of telophase cells to reorient in response to apical cues (Figure 7F; Figure 7—figure supplement 2C,D).

These findings, together with our observation that AJ and LGN mutants differ in their planar telophase correction phenotypes, suggest that afadin, α-E-catenin and vinculin act independently of LGN in the context of spindle orientation. As further evidence, we find that while LGN strongly colocalizes with known binding partners Gαi3 and Insc, afadin demonstrates minimal colocalization with LGN either pre- or post-chromosome segregation (Figure 7—figure supplement 2E). Finally, there are several contexts during epidermal development where LGN is not required for division orientation. First, although hair placode progenitors undergo perpendicular asymmetric divisions (Ouspenskaia et al., 2016), LGN is weakly expressed and is not required for proper division orientation in mitotic placode cells (Byrd et al., 2016). Second, while LGN loss reduces perpendicular divisions in the interfollicular epidermis at E16.5, LGN is dispensable at E14.5, when the majority of divisions are planar and LGN is rarely cortical (Williams et al., 2014). Conversely, Afdn knockdown increases the frequency of oblique divisions in both contexts, suggesting an LGN-independent function for afadin in both perpendicular and planar divisions (Figure 7—figure supplement 2F,G). Together, these data suggest that afadin is a minor or transient LGN-interactor in vivo and support a polarity- and LGN-independent role for afadin in telophase correction.

Telophase correction and early mitotic spindle orientation function as parallel pathways

We next sought to test genetically whether the telophase correction pathway can override the initial spindle positioning cues provided by LGN. To address this, we generated afadin and LGN dual loss-of-function embryos by injecting the Gpsm21617 lentivirus into either a wild-type (Afdnfl/fl) or Afdn cKO background. While loss of LGN alone recapitulated the previously described phenotype of impaired stratification (Williams et al., 2011), loss of Afdn on an Gpsm2 mutant background partially rescued this differentiation defect (Figure 7G,H). Moreover, the predominantly planar division orientation observed in Gpsm2 single mutants became more randomized upon dual loss with Afdn, generating an intermediate phenotype (Figure 7I). These epistasis experiments suggest that telophase correction operates in parallel with, rather than downstream of, the canonical spindle orientation pathway.

Of note, double mutants largely lacked perpendicular (70°−90°) divisions, further supporting a specific role for the LGN complex to generate this division type. Taken together, these data suggest that early spindle orientation cues direct imprecise perpendicular divisions in an LGN-dependent manner. These divisions are then refined into the characteristic bimodal pattern of perpendicular or planar divisions by telophase correction. However, these data also suggest that LGN-directed perpendicular correction is still dependent on the AJ components driving telophase correction.

Telophase correction also occurs during early stratification

The observations that afadin is required for telophase correction at E16.5 (Figure 6F-J), and that Afdn mutants display division orientation defects at both early and peak stages of stratification (Figure 5G-I; Figure 7—figure supplement 2G) prompted us to examine whether telophase correction occurs throughout epidermal morphogenesis. Thus, we performed live imaging on wild-type Krt14Cre; Rosa26mT/mG epidermal explants at E14.5, when stratification initiates (Figure 8A). Even though nearly all divisions at E14.5 are planar (Lechler and Fuchs, 2005; Williams et al., 2014), remarkably, in wild-type cells at telophase onset, the distribution of observed orientations was randomly-distributed, similar to what was observed at E16.5 (Figure 8B; compare to Figure 1G). However, while 47% of cells (n=78) entered telophase oriented obliquely, the vast majority of these possessed a basal endfoot and corrected to planar within 1h of telophase onset (Figure 8C). On the other hand, the few cells (28%) that did not maintain basal contact corrected randomly at E14.5, in contrast to E16.5, when they invariably corrected to perpendicular (compare Figure 8C to 3B). Since LGN does not localize cortically or influence division orientation at E14.5 (Williams et al., 2014), this provides additional evidence that LGN is necessary for perpendicular telophase correction.

Figure 8 with 1 supplement see all
Planar telophase correction limits precocious and sustained differentiation and biases clones towards symmetric cell divisions.

(A) (top) z-projection stills from a movie of an E14.5 mitotic cell, annotated as in Figure 1E. (B) Cumulative frequency distribution of division angles from live imaging experiments of E14.5 embryos at anaphase onset (blue; φ) and 1h later (black; θ); n indicates number of divisions from 3 embryos across two independent sessions. (C) Data from (B) depicting division orientations at telophase and 1h later, sorted based on retention/loss of basal contact throughout cell division. Connecting lines demonstrate that, at E14.5, planar correction occurs in a contact dependent manner, while mitoses that lose contact demonstrate no obvious pattern of correction. (D) Sagittal section of E14.5 epidermis with mosaic Afdn2711 H2B-RFP transduction. Regions of high infection display increased stratification, as demonstrated by K10 (green) positivity. (E-F) Quantification of epidermal differentiation from E14.5 sagittal sections. Afdn (E) or Ctnna1 (F) knockdown increases suprabasal cell density, suggesting precocious differentiation. (G) (top) Graphical depiction of clonal lineage tracing strategy; (bottom) Representative images of E17.5 sagittal sections from lineage tracing experiments stained with GFP (green), K10 (red), and RFP/β4-integrin (gray). Afdn2711 knockdown clones display asymmetric (suprabasal) bias. (H) Clonal density arrays representing all evaluated clones (except delamination events) from experiments outlined in (G). The proportion of total clones for each possible combination of basal/suprabasal cells is coded on a color spectrum correlating to 0–12% of all clones. (I) Quantification of suprabasal (SB):basal cell ratio for individual clones. Knockdown of Afdn results in a higher ratio of SB cells in individual clones compared to wild-type littermates. Scale bars, 5 μm (A), 25 μm (D). *p<0.05, ***p<0.001, determined by Kolmogorov-Smirnov test (B) or student’s t-test (F). See also Figure 8—figure supplement 1.

Telophase correction impacts cell fate decisions

At E14.5, Afdn mutants displayed fewer planar and more oblique divisions compared to controls (Figure 7—figure supplement 2G), which led us to ask whether afadin loss could promote precocious differentiation. In E14.5 Afdn2711 mosaic epidermis, we noted that Keratin-10 (K10)—a marker of differentiated cells—was enriched in RFP+ mutant regions compared to RFP- wild-type regions (Figure 8D). While basal cell density was similar between Afdn2711 embryos and non-transduced littermates, the density of differentiated cells—whether assessed by their suprabasal (SB) position or K10 expression—was significantly higher in Afdn mutants (Figure 8E, Figure 8—figure supplement 1A). This was unlikely to be caused by hyperproliferation because similar levels of mitotic cells were observed at both E14.5 and E16.5 in Afdn2711 RFP+ and wild-type littermate controls (Figure 8—figure supplement 1B). Like afadin, loss of α-E-catenin resulted in a hyperstratified epidermis and increased suprabasal cell density (Figure 8F; Figure 8—figure supplement 1C). Consistent with previous observations in E18.5 Ctnna1 knockout epidermis (Beronja et al., 2010; Vasioukhin et al., 2001), the precocious differentiation observed in these mutants persisted into later ages (Figure 8—figure supplement 1D, E). Notably, in contrast to a previous report that late embryonic Ctnna1 epidermis is hyperproliferative (Vasioukhin et al., 2001), we do not observe any elevation in mitotic cells in either E14.5 or E16.5 Ctnna1 epidermis (Figure 8—figure supplement 1F), which is more in agreement with a recent study that showed a mild increase in BrdU+ cells but net growth disadvantage of Ctnna1912 basal cells (Beronja et al., 2010). Thus, we feel it is more likely that the precocious differentiation observed in Ctnna1912 mutants is due to persistent oblique divisions caused by errors in telophase correction, rather than to hyperproliferation. Collectively, these data suggest that telophase correction influences differentiation throughout epidermal development.

While the previous experiments demonstrated that AJ loss alters both division orientation and promotes differentiation, they do not address whether telophase correction errors directly impact cell fate choices. To explore whether afadin loss alters fate decisions, we performed short term (72h) lineage tracing experiments using Krt14CreER; Rosa26Confetti reporter mice in combination with Afdn2711 knockdown, and examined the number of progenitor (basal) and differentiated (SB, K10+) progeny within resultant clones (Figure 8G). The lentiviral shRNA strategy was chosen to target Afdn because the alternative—Krt14CreER-mediated deletion of the Afdnfl allele—would result in Afdn deletion occurring simultaneous with, rather than prior to, clonal induction. Moreover, in contrast to lentiviral-delivered Afdn2711, Krt14Cre-mediated deletion of Afdn did not cause obvious differentiation defects (compare Figure 7H to Figure 8—figure supplement 1E), likely because Afdn deletion occurs later with Krt14Cre (Beronja et al., 2010).

We administered a single dose of tamoxifen at E14.5 by oral gavage, then harvested embryos at E17.5, when we analyzed clones obtained from Afdn2711 RFP+ and uninjected (wild-type) littermates. In agreement with our mitotic index measurements, knockdown of Afdn did not alter the distribution of clone sizes (Figure 8—figure supplement 1G). However, Afdn2711 clones frequently displayed a greater proportion of suprabasal cells per clone when compared to uninjected littermates (Figure 8; Figure 8—figure supplement 1H). We utilized clonal density arrays to display clone size distributions, such that basal and SB cells/clone are plotted on x and y axes, respectively, and darker colors indicate higher frequencies of specific clone types (Byrd et al., 2019). These data demonstrate that Afdn2711 contain a higher proportion of SB-rich clones (Figure 8I).

We further characterized clones into four subtypes: (1) balanced (1:1 ratio of basal:SB cells), (2) basal-rich (basal:SB ratio>1), (3) SB-rich (basal:SB ratio<1), and delamination (basal cells=0). Delamination is an alternative differentiation mechanism to asymmetric cell division, whereby a basal cell detaches from the underlying basement membrane and initiates differentiation without dividing. We previously showed through lineage tracing that delamination drives the initial phase of stratification, while asymmetric cell divisions predominate during peak stratification (Williams et al., 2014). While we did not observe clear delamination events in our 3-6 hour live imaging experiments, genetic lineage tracing revealed that a similar and significant fraction of clones in both WT and Afdn2711 epidermis (48% vs 40%) arose from delamination (Figure 8—figure supplement 1I). A comparison of the mitotic clone distribution between WT and Afdn2711 clones revealed that Afdn2711 epidermis contains a much greater number of SB-rich clones (24% vs 2% in WT), at the expense of the basal-rich (29% vs 41%) subtype (Figure 8—figure supplement 1I). Since delamination events slightly decrease in Afdn mutants compared to WT controls, this further suggests that the excess differentiation observed in Afdn mutants is attributable to an increase in asymmetric cell divisions rather than compensatory delamination.

We conclude that the excess oblique divisions observed in Afdn mutants, which fail to be corrected during telophase—impacts cell fate decisions, favoring differentiation over self-renewal. This further implies that a high proportion of oblique divisions are operationally asymmetric. In conclusion, we provide several lines of evidence that telophase correction contributes to establishing proper epidermal architecture: (1) the tensile AJ components afadin, α-E-catenin and vinculin fail to correct during telophase, leading to a persistent excess of oblique divisions, (2) AJ mutants which fail at telophase correction induce excess stratification, and (3) the failure of oblique divisions to correct to planar during telophase leads to a bias toward differentiation over self-renewal.

Discussion

A two-step mechanism for division axis determination

These studies shed new light on the mechanisms governing oriented cell divisions in the developing epidermis and identify telophase correction as an important contributor to balancing symmetric and asymmetric divisions throughout stratification. While previous studies have demonstrated essential roles for canonical spindle orientation genes in division orientation, we now show that initial spindle positioning is only one part of the process (Figure 9). Our data suggest that LGN and associated proteins operate early in mitosis to promote perpendicular divisions, but do so with a high degree of imprecision, resulting in a wide distribution of anaphase division angles. While this function of LGN is required for perpendicular divisions to occur, this fails to explain the bimodal distribution of division angles observed in telophase. In the second phase of our model, telophase cells undergo dynamic reorientation towards a planar or perpendicular orientation, where the direction of correction is dependent on contact with the basement membrane via a basal endfoot. We further demonstrate that LGN is also required for this second phase of spindle orientation, as its maintenance at the apical cortex promotes perpendicular-directed telophase correction. Moreover, the fidelity of telophase correction relies on the actin-scaffolding α-E-catenin/vinculin/afadin pathway, highlighting a role for cell adhesion and cytoskeletal dynamics in division orientation. In this way, our findings now provide a mechanistic explanation for the randomized division orientation observed in Ctnna1 mutants more than a decade ago (Lechler and Fuchs, 2005). Importantly, while our data support a model wherein vinculin regulates dynamic assembly of AJs, we cannot exclude the possibility that vinculin may play similar roles in cell-matrix integrin adhesions, which may also impact telophase correction.

Two-step model of division orientation.

Model of OCD in the embryonic epidermis. During stratification, LGN (green) is recruited to the apical cortex in ~50% of mitoses, promoting perpendicular divisions. For OCDs with perpendicular and planar anaphase orientations, the division angle is fixed at anaphase onset, exhibiting minimal change in radial orientation during telophase. Importantly, the activity of LGN and its binding partners is imprecise, frequently resulting in oblique orientations at anaphase. In these cases, the apical daughter either retains or loses basement membrane contact following cytokinesis (red or blue nuclei, respectively). If contact is maintained, the apical daughter will reorient into a planar position. In contrast, if contact is lost, the apical daughter further stacks above its basal partner. Upon loss of α-E-catenin, vinculin, or afadin, telophase reorientation in either direction fails, resulting in persistent oblique divisions. In comparison, LGN loss reduces perpendicular anaphase orientations, while oblique divisions are properly corrected in a contact dependent manner. Afdn loss on an Gpsm2 mutant background restores oblique divisions and largely rescues the Gpsm2 differentiation defect.

Corrective mechanisms in oriented cell divisions

Our findings contribute to a growing number of corrective mechanisms which can counterbalance stem cell division orientation errors in order to preserve tissue homeostasis. In Drosophila neuroblasts, the "telophase rescue" pathway—mediated by the scaffolding protein Dlg and motor protein Khc73—can compensate for errors in spindle orientation by relocalizing fate determinants, thus preserving normal daughter cell fates (Cai et al., 2001; Peng et al., 2000; Siegrist and Doe, 2005). However, telophase rescue differs from the telophase correction we report here in that division orientation errors are not corrected in telophase, but rather, the fate determinants themselves are repositioned relative to the new division axis. In the developing epidermis, it has been shown that Insc overexpression can promote apical LGN localization and drive an increase of perpendicular divisions (Poulson and Lechler, 2010; Williams et al., 2011), but that under some circumstances, NuMA can redistribute laterally, perhaps in an effort to prevent the hyper-differentiation that would be driven by excessive asymmetric divisions (Poulson and Lechler, 2010). Our data here, where afadin, α-E-catenin, and vinculin can override the perpendicular-correcting cue provided by LGN, provide a potential molecular explanation for this plasticity.

Other examples of dynamic oriented cell divisions include cyst stem cells of the Drosophila testis, which display randomized spindle angles until anaphase, at which point one centrosome becomes anchored at the interface with the niche-defining hub cell, driving division away from the niche (Cheng et al., 2011). In addition, dividing cells within the monolayered Drosophila follicular epidermis partially extrude during mitosis and frequently demonstrate oblique division angles, which are corrected by reinsertion into the epithelium in an adhesion-dependent manner (Bergstralh et al., 2015). A more extreme example of this extrusion/reinsertion model has been observed in intestinal organoids, where mitotic intestinal stem cells migrate to the luminal surface and undergo planar divisions before reinserting into the epithelium on either side of a Paneth cell (McKinley et al., 2018). Furthermore, genetic alterations in MDCK cells—specifically, Gpsm2 knockdown or Par1b overexpression—can drive out-of-plane divisions which are capable of correcting during anaphase via an apical actomyosin compressive force (Lázaro-Diéguez and Müsch, 2017; Zheng et al., 2010). Taken together, these studies and ours suggest that many of these corrective mechanisms rely on polarity, cell-adhesion, and actin dynamics.

Insights into epidermal cell fate specification

In the Drosphila neuroblast, the orientation of cell division is directly linked to cell fates via the asymmetric inheritance of transcription factors and other fate determinants which promote differentiation in one daughter cell and preserve stemness in the other (Bergstralh et al., 2017; Knoblich, 2008). While no such fate determinant has been identified in epidermal progenitors, our results add to a growing body of evidence that division orientation and cell fates are tightly linked. While previous studies have used short-term lineage tracing to correlate patterns of division orientation with fate choices (Poulson and Lechler, 2010; Williams et al., 2014), the lineage tracing experiments performed in this study are the first to demonstrate that perturbations to division orientation lead to altered cell fate outcomes.

Importantly, given the timing of telophase correction, our observations also shed new light on the timing and speculative mechanisms of cell fate commitment during mitosis. In the normal developing epidermis, a large proportion of mitoses (30–40%) progress to anaphase at oblique orientations. While telophase reorientation normally sorts these indeterminate divisions into symmetric or asymmetric outcomes, the evidence from loss-of-function experiments in Ctnna1 and Afdn mutants—resulting in a hyper-stratified epidermis and lineages biased toward differentiation—suggests that a significant portion of oblique divisions are operationally asymmetric, likely resulting in differentiation of the obliquely-positioned daughter cell. Furthermore, these data suggest that retention of basement membrane contact is a potentially potent driver of basal progenitor identity. Taken together, our results indicate that while the early presence of LGN in pro/metaphase may bias cells towards adopting an asymmetric outcome, the finality of this decision is not determined until telophase reorientation mechanisms push or pull cells into the suprabasal or basal layers, respectively. Telophase correction thus provides a potential source of plasticity in the fate choices made by epidermal basal cells. It is tempting to speculate that AJ components in mitotic cells function as a mechanosensor that tranduces information about the local cellular environment that favors planar correction when tension is high and perpendicular correction when tension is low.

Materials and methods

Key resources table
Reagent type
(species) or resource
DesignationSource or ref.IdentifiersAdditional
information
Strain, strain background
(Mus musculus)
mTmGJackson LabsIMSR Cat# JAX:007576, RRID:IMSR_JAX:007576
Strain, strain background
(Mus musculus)
Krt14Cre(Dassule et al., 2000)
Strain, strain background
(Mus musculus)
Krt14H2B-GFP(Tumbar et al., 2004)
Strain, strain background
(Mus musculus)
CD1Charles RiverIMSR Cat# CRL:022, RRID:IMSR_CRL:022New females integrated into colony every ~ 5 years to maintain as outbred strain.
Strain, strain background
(Mus musculus)
LSL-ConfettiJackson LabsIMSR Cat# JAX:013731, RRID:IMSR_JAX:013731
Strain, strain background
(Mus musculus)
Krt14CreERJackson LabsIMSR Cat# JAX:005107, RRID:IMSR_JAX:005107
Strain, strain background
(Mus musculus)
Afdnfl/fl(Beaudoin et al., 2012)
AntibodyGuinea-pig polyclonal anti-LGN(Williams et al., 2011)(1:500 dilution)
AntibodyRabbit monoclonal anti-survivin (71G4B7)Cell SignalingCat# 2808, RRID:AB_2063948(1:1000 dilution)
AntibodyChicken polyclonal anti-GFPAbcamCat# ab13970, RRID:AB_300798(1:2000 dilution)
AntibodyRat monoclonal anti-mCherry (16D7)Thermo Fisher ScientificCat# M11217, RRID:AB_2536611(1:2000 dilution)
AntibodyRabbit polyclonal anti-LGNMilliporeCat# ABT174(1:2000 dilution)
AntibodyRat monoclonal anti-β4 integrinThermo-FisherBD Biosciences Cat# 553745, RRID:AB_395027(1:1000 dilution)
AntibodyRabbit polyclonal anti-Gai3EMD MilliporeMillipore Cat# 371726–50 UL, RRID:AB_211897(1:500 dilution)
AntibodyGoat polyclonal anti-dynactinAbcamAbcam Cat# ab11806, RRID:AB_298590(1:500 dilution)
AntibodyMouse (IgM) monoclonal anti-NuMABD Transduction LabsBD Biosciences Cat# 610562, RRID:AB_397914(1:300 dilution; use Jackson labs Donkey anti-IgM Cy3 conjugated secondary)
AntibodyRabbit polyclonal anti-histone H3, phospho (Ser10)MilliporeCat# 06–570, RRID:AB_310177(1:1000 dilution)
AntibodyRat monoclonal anti-a-tubulinEMD MilliporeMillipore Cat# CBL270, RRID:AB_93477(1:500 dilution)
AntibodyRabbit polyclonal anti-pericentrinCovanceCovance Cat# PRB-432C-200, RRID:AB_291635(1:500 dilution)
AntibodyRabbit polyclonal anti-cytokeratin 10 (Poly19054)Bio-LegendCat# 905404, RRID:AB_2616955(1:1000 dilution)
AntibodyRabbit polyclonal anti-Par3EMD MilliporeMillipore Cat# 07–330, RRID:AB_2101325(1:500 dilution)
AntibodyRat monoclonal anti-E-cadherin (ECCD-2)Life TechnologiesThermo Fisher Scientific Cat# 13–1900, RRID:AB_2533005(1:1000 dilution)
AntibodyGoat polyclonal anti-E-cadherinR and D systemsR and D Systems Cat# AF748, RRID:AB_355568(1:1000 dilution)
AntibodyRabbit polyclonal anti-a-E-cateninInvitrogenThermo Fisher Scientific Cat# 71–1200, RRID:AB_2533974(1:300 dilution; tissue sections)
AntibodyRat monoclonal anti-a18(Yonemura et al., 2010)(1:10000 dilution; tissue sections) (1:2000 dilution; wholemounts)
AntibodyRabbit polyclonal anti-vinculinGift from Dr. Keith Burridge(1:1000 dilution)
AntibodyMouse monoclonal anti-vinculinSigma AldrichSigma-Aldrich Cat# V9131, RRID:AB_477629(1:500 dilution)
AntibodyRabbit polyclonal anti-afadinSigma AldrichSigma-Aldrich Cat# A0224, RRID:AB_257871(1:500 dilution)
AntibodyMouse monoclonal anti-pMLC2 (Ser19)Cell SignalingCell Signaling Technology Cat# 3675, RRID:AB_2250969(1:500 dilution)
Chemical compoundPhalloidin AF-647 conjugatedLife TechnologiesThermo Fisher Scientific Cat# A22287, RRID:AB_2620155(1:500 dilution)
Cell line(s)
(Mus musculus)
Primary keratinocytesThis publication.Isolated as described in Materials and methods section.
Chemical compoundTamoxifenSigma-AldrichCat# T5648
SoftwareFIJISource:https://imagej.net/Fiji
Reference: doi:10.1038/nmeth.2019

Animals

Mice were housed in an AAALAC-accredited (#329; June 2017), USDA registered (55-R-0004), NIH welfare-assured (D16-00256 (A3410-01)) animal facility. All procedures were performed under IACUC-approved animal protocols (16-162). For live imaging experiments we utilized either: (1) mT/mG (Gt(ROSA)26Sortm4(ACTB-tdTomato,-EGFP)Luo/J; Jackson Labs #007576 via Liqun Luo, Stanford University) homozygous females with at least one copy of the Krt14Cre allele (Dassule et al., 2000) (crossed to males of the identical genotype), or (2) Krt14H2B-GFP (Tumbar et al., 2004) and Rosa26mT/mG heterozygous females (crossed to identical males). For lineage tracing experiments (see below for additional details) we crossed Krt14CreER; Rosa26Confetti females to identical males (Tg(KRT14-cre/ERT)20Efu; Jackson Labs #005107/Gt(ROSA)26Sortm1(CAG-Brainbow2.1)Cle; Jackson Labs #013731). For fixed sample imaging, wild-type CD1 mice (Charles River; #022) were utilized. Afdnfl/fl animals (Beaudoin et al., 2012) were maintained on a mixed C57B6/J CD1 background and either bred to the same Krt14-Cre allele or injected with lentiviral Cre-mRFP1 (see below). The procedure for producing, concentrating and injecting lentivirus into amniotic fluid of E9.5 embryos has been previously described and is briefly detailed below (Beronja et al., 2010).

Live imaging

Request a detailed protocol

The live imaging protocol used in this study was adapted from the technique recently described by the Devenport lab (Cetera et al., 2018). A 1% agar solution/media solution containing F-media (3:1 DMEM:F12 + 10% FBS + 1% Sodium bicarbonate + 1% Sodium Pyruvate + 1% Pen/Strep/L-glut mix), was cooled and cut into 35mm discs. Epidermal samples measuring ~4-6mm along the AP axis and ~2-3mm along the medial-lateral axis were extracted from the mid-back of E16.5 mT/mG embryos. These explants were placed dermal-side down onto the gel/media disc, then sandwiched between the gas-permeable membrane of a 35mm lumox culture dish (Sardstedt; 94.6077.331). Confocal imaging was performed utilizing a Zeiss LSM 710 Spectral confocal laser scanning microscope equipped with a 40X/1.3 NA Oil Plan Neo objective. Images were acquired with 5 minute intervals and a Z-series with 0.5 µm step-size (total depth ranging from 20-30 microns) for 3-9 hours. Explants were cultured at 37°C with 5.0% CO2 for >1.5 hours prior to- and throughout the course of imaging. Divisions occurring close to the tissue edge or showing any signs of disorganization/damage were avoided to exclude morphological changes associated with wound-repair. 4D image sets were deconvolved using AutoQuant X3 and processed using ImageJ (Fiji).

Lentiviral injections

Request a detailed protocol

For full protocol, please see Beronja, et al. (ref. 24). This protocol is approved via IACUC #16-162/19-155. Pregnant CD1, mTmG/Krt14-Cre, or Afdnfl/fl females were anesthetized and the uterine horn pulled into a PBS filled dish to expose the E9.5 embryos. Embryos and custom glass needles were visualized by ultrasound (Vevo 2100) to guide microinjection of ~0.7 μl of concentrated lentivirus into the amniotic space. Three to ten embryos were injected depending on viability and litter size. Following injection, the uterine horn(s) were reinserted into the mother’s thoracic cavity, which was sutured closed. The incision in the skin was resealed with surgical staples and the mother provided subcutaneous analgesics (5 mg/kg meloxicam and 1-4 mg/kg bupivacaine). Once awake and freely moving, the mother was returned to its housing facility for 5-7 days, at which point E14.5-16.5 embryos were harvested and processed accordingly.

Lineage tracing

Request a detailed protocol

Krt14CreER; Rosa26Confetti females were mated to males with the identical genotype. At E9.5, ~half of the viable embryos were injected with Afdn2711 H2B-mRFP1 high titer lentivirus (see above for detailed surgical procedure). Activation of the Krt14CreER allele was initiated by tamoxifen (dosed at 100 μg per gram dam mass) delivered by oral gavage at E14.5, five days following lentiviral injection). Females were monitored for 24 hours following tamoxifen dosing for signs of abortion or distress. Embryos were harvested at E17.5 (~72 hours after tamoxifen delivery) and backskins were embedded in OCT and sectioned sagittally (8μm thick sections). Slides were stained with Abcam Chicken αGFP polyclonal antibody (Abcam ab13970) which enhanced the membrane-CFP, nuclear-GFP, and cytoplasmic-YFP fluorophores of the Confetti allele. Images were acquired for every labeled clone using a 40x/1.15NA objective with a 1.5X digital zoom. Sparse clones (<1% total cells) were evaluated for both the number of basal and suprabasal cells (distinguished by staining with αKrt10 antibody; Figure 8I). Clones with only suprabasal cells in the stratum spinosum or first stratum granulosum (SG3) layer were assumed to be delamination events – those above SG3 were excluded. Suprabasal (SB) to basal cell ratios were quantified for each clone by dividing the # of SB cells by the # of basal cells. Clones with a ratio >1 were binned as 'SB-rich' while clones with a ratio <1 were binned as 'basal-rich' – clones with an equal number (ratio = 1) were binned as 'balanced'.

Constructs and RNAi

Request a detailed protocol

For afadin and vinculin RNAi targeting, we tested ~10 shRNAs for knockdown efficiency in primary keratinocytes. These sequences were selected from The RNAi Consortium (TRC) Mission shRNA library (Sigma) versions 1.0, 1.5, and 2.0 and cloned using complementary annealed oligonucleotides with AgeI/EcoRI linkers. For LGN and α-catenin, we utilized an shRNA that had been previously validated with our lentiviral injection technique (Beronja et al., 2010; Williams et al., 2011). shRNA clones are identified by the gene name with the nucleotide base (NCBI Accession number) where the 21-nucleotide target sequence begins in superscript (e.g. Afdn2711). Lentivirus was packaged in 293FT or TN cells using the pMD2.G and psPAX2 helper plasmids (Addgene plasmids #12259 and #12260, respectively). For knockdown screening, primary keratinocytes were seeded at a density of ~150,000 cells per well into 6-well plates and grown to ~80% confluency in E-Low calcium medium and infected with an MOI of ~1. Approximately 48 h post-infection, keratinocytes were treated with puromycin (2 μg/mL) to generate stable cell lines. After 3-4 days of puromycin selection, cells were lysed and RNA isolated using the RNeasy Mini Kit (Qiagen). cDNA was generated and amplified from 10-200 μg total RNA using either Superscript VILO (Invitrogen) or iScript (Bio-Rad). mRNA knockdown was determined by RT-qPCR (Applied Biosystems 7500 Fast RT-PCR) using 2 independent primer sets for each transcript with Hprt1 and cyclophilin B (Ppib2) as reference genes and cDNA from stable cell lines expressing Scramble shRNA as a reference control. Primer efficiencies were determined using dose-response curves and required to be >1.8, with relative transcript abundance determined by the ΔΔCT method. RT-qPCR runs were performed in triplicate with the mean knockdown efficiency determined by calculating the geometric mean of the ΔΔCT values for at least two independent technical replicates. The following primer sequences were used: Afdn (fwd-1: 5’- ACGCCATTCCTGCCAAGAAG -3’, rev-1: 5’- GCAAAGTCTGCGGTATCGGTAGTA -3’; fwd-2: 5’- GGGGATGACAGGCTGATGAAA -3’, rev-2: 5’- CGATGCCGCTCAAGTTGGTA -3’), Vcl (fwd-1: 5’- TACCAAGCGGGCACTTATTCAGT -3’, rev-1: 5’- TTGGTCCGGCCCAGCATA -3’; fwd-2: 5’- AAGGCTGTGGCTGGAAACATCT -3’, rev-2: 5’- GGCGGCCATCATCATTGG -3’). The following shRNA targeting sequences were used: Afdn2711 (5’- CCTGATGACATTCCAAATATA -3’), Vcl3466 (5’- CCCTGTACTTTCAGTTACTAT -3’), Vcl2803 (5’- CCACGATGAAGCTCGGAAATG -3’), Ctnna1912 (5’-CGCTCTCAACAACTTTGATAA -3’), Gpsm21617 (5’- GCCGAATTGGAACAGTGAAAT -3’), Scramble (5’- CAACAAGATGAAGAGCACCAA-3’).

Antibodies, immunohistochemistry, and fixed imaging

Request a detailed protocol

E14.5 embryos were mounted whole in OCT (Tissue Tek) and frozen fresh at -20°C. E16.5 embryos were skinned and flat-mounted on Whatman paper. In both cases, infected and uninfected littermate controls were mounted in the same blocks to allow for direct comparisons on the same slide. For α-tubulin staining of metaphase spindles, samples were kept warm and pre-fixed with room-temperature 4% paraformaldehyde for 10 minutes before OCT embedding. Frozen samples were sectioned (8 μm thick) on a Leica CM1950 cryostat, mounted on SuperFrost Plus slides (ThermoFisher) and stored at -80°C. For staining, sections were thawed at 37°C for 5-15 min, fixed for 5 min with 4% paraformaldehyde, washed with PBS and blocked for 1h with gelatin block (5% NDS, 3% BSA, 8% cold-water fish gelatin, 0.05% Triton X-100 in PBS). Primary antibodies were diluted in gelatin block and incubated overnight in a humidity chamber at 4°C. Slides were then washed with PBS and incubated with secondary antibodies diluted in gelatin block at room temperature (~25°C) for 2 hours, counterstained with DAPI (1:2000) for 5 minutes and mounted in ProLong Gold (Invitrogen). Actin was visualized by phalloidin-AF647 staining (Life Technologies; 1:500) simultaneously with secondary antibody incubation. Images were acquired using LAS AF software on a Leica TCS SPE-II 4 laser confocal system on a DM5500 microscope with ACS Apochromat 20x/0.60 multi-immersion, ACS Apochromat 40x/1.15 oil, or ACS Apochromat 63x/1.30 oil objectives.

The following primary antibodies were used: survivin (rabbit, Cell Signaling 2808S, 1:500), LGN (Williams et al., 2011) (guinea pig, 1:500), LGN (rabbit, Millipore ABT174, 1:2000), phospho-histone H3 (rat, Abcam ab10543, 1:1,000), mCherry (rat, Life Technologies M11217, 1:1000-3000), β4-integrin (rat, ThermoFisher 553745, 1:1,000), Gαi3 (rabbit, EMD Millipore 371726, 1:500), GFP (chicken, Abcam ab13970, 1:1,000), dynactin (goat, Abcam ab11806, 1:500), NuMA (mouse IgM, BD Transduction Labs 610562, 1:300), α-tubulin (rat, EMD Millipore CBL270, 1:500), pericentrin (rabbit, Covance PRB-432C, 1:500), Par3 (rabbit, EMD Millipore 07-330, 1:500), E-cadherin (rat, Life Technologies 131900, 1:1,000), E-cadherin (goat, R&D System AF748, 1:1,000), α-E-catenin (rabbit, Invitrogen 71-1200, 1:300) α18 (rat, generous gift of Dr. Nagafuchi at Nara Medical University, 1:10,000), vinculin (mouse IgG, Sigma V9131, 1:500), vinculin (rabbit, generous gift of Dr. Keith Burridge at University of North Carolina, 1:1000), afadin (rabbit, Sigma A0224, 1:500), pMLC2 (Ser19) (mouse IgG, Cell Signaling 3675S, 1:1000). Actin labeling achieved via Phalloidin-AF647 (Life Technologies A22287, 1:500) in secondary antibodies

The following secondary antibodies were used (all antibodies produced in donkey): anti-rabbit AlexaFluor 488 (Life Technologies, 1:1000), anti-rabbit Rhodamine Red-X (Jackson Labs, 1:500), anti-rabbit Cy5 (Jackson Labs, 1:400), anti-rat AlexaFluor 488 (Life Technologies, 1:1000), anti-rat Rhodamine Red-X (Jackson Labs, 1:500), anti-rat Cy5 (Jackson Labs, 1:400), anti-guinea pig AlexaFluor 488 (Life Technologies, 1:1000), anti-guinea pig Rhodamine Red-X (Jackson Labs, 1:500), anti-guinea pig Cy5 (Jackson Labs, 1:400), anti-goat AlexaFluor 488 (Life Technologies, 1:1000), anti-goat Cy5 (Jackson Labs, 1:400), anti-mouse IgG AlexaFluor 488 (Life Technologies, 1:1000), anti-mouse IgG Cy5 (Jackson Labs, 1:400), anti-mouse IgM Cy3 (Jackson Labs, 1:500).

Keratinocyte culture and Calcium-shift assays

Request a detailed protocol

Primary mouse keratinocytes were maintained in E medium with 15% chelated FBS and 50 μM CaCl2 (E low medium). For viral infection, keratinocytes were plated at ~150,000 cells per well in a 6-well plate and incubated with lentivirus in the presence of polybrene (1 μg/mL) and centrifuged at 1,100 xg for 30 min at 37°C. All shRNA cell lines were derived from the same wild-type lineage (primary CD1 mouse keratinocytes isolated from P3 backskins). Stable cell lines were generated/maintained by adding puromycin (2 μg/mL) 48 h after infection and continual antibiotic treatment following. The Afdnfl/fl and Afdnfl/fl; Krt14Cre (Afdn-cKO) keratinocyte lines were isolated from P3 littermates and used at low passage (<P10). Cell line identity was doubly confirmed by knockdown/knockout specificity via immunofluorescent staining. All lines tested negative for mycoplasma using the ATCC 30-1012K kit. Calcium shifts were performed by seeding ~45,000 low passage cells (<P10) per well into 8-well Permanox chamber slides (Lab-Tek 177445) coated with poly-L-lysine, collagen, and fibronectin. Once cells reached ~85% confluency (~12-16 hours) cells were switch to high Ca2+ (1.5mM) medium and grown for the indicated period of time (30 min to 8 hours). Cells were fixed with 4% paraformaldehyde in PBS warmed to room-temperature. Immunostaining was performed using the same protocol as for tissue sections (see above).

3T3 fibroblasts and HEK-293 cells – both of which are included on the International Cell Line Authentication Committee’s register of misidentified cell lines (version 9) were specifically used in primary keratinocyte isolation and lentiviral production, respectively. Neither cell line was utilized in any experimental procedures.

Measurements, quantification, graphing, and statistics

Spindle and division orientation

Request a detailed protocol

Mitotic cells in metaphase were identified based on nuclear morphology. Metaphase spindle orientation was measured as the angle between a vector orthogonal to the metaphase plate and parallel to the basement membrane. Anaphase cells were identified by both nuclear condensation and widely distributed surviving staining between daughter cells. Telophase cells were distinguished due to reduced nuclear condensation and dual-punctate Survivin staining. Division orientation was measured as the angle between a vector connecting the center of each daughter nucleus and a vector running parallel to the basement membrane. The same methodology was used to measure division orientation in live imaging experiments. In cases lacking nuclear labeling, the position of the nuclei was inferred based on cell volume/shape changes. Telophase correction (θ-ϕ) was quantified as the difference between division orientation at anaphase onset (ϕ) and division orientation 1 h later (θ). The presence of basal contact for the more apical daughter was determined by analyzing cell morphology in both en face and orthogonal perspectives.

Adhesion assays

Request a detailed protocol

Quantification of fluorescence intensity in adhesion assays was performed by orthogonal linescans at three positions along the junction length (~25th, 50th, and 75th quarter). In cases where junctions appeared punctate, discreet puncta were evaluated to avoid measuring regions lacking junction formation. Signal centers were set based on maximum intensity of either E-cadherin or α-E-catenin (where appropriate). To quantify ratios, the geometric mean fluorescent intensities of the 3 values nearest the junction center were used. Quantification of junction continuity was performed by linescans of E-cadherin fluorescence intensity along the entire length of the junction, excluding the vertex of multiple cells (i.e. tricellular junctions). We then calculated % of these intensity measurements above a threshold, which was evaluated for each individual junction using the mean center intensity of three orthogonal scans described earlier in this paragraph.

LGN localization/intensity

Request a detailed protocol

LGN localization patters (e.g. apical, weak/absent, or other) were determined for cells labeled with pHH3, irrespective of the lentiviral H2B-RFP reporter to avoid bias. Imaging was performed with WT controls and experimental samples on the same slide to avoid variation in antibody staining. Radial localization of LGN was measured by determining the angle between two vectors: one drawn from the LGN signal center to the center of the nucleus, the other drawn parallel to the basement membrane. Crescents oriented at the apical side were given positive values, while those at the basal side were given negative values. Radial variance between LGN signal and spindle or division axis were determined by drawing two vectors: one for the radial orientation of the LGN signal center and a second between either the spindle poles or between the center of the daughter nuclei. Radial fluorescent intensity values were measured by linescans originating at the site of basement membrane contact and tracing the edge of the cell. Each measurement along the length of the scan was then set as a part of whole, operating with the assumption that ~50% of the total length would represent the apical surface.

Whole mount fluorescence intensity

Request a detailed protocol

Quantification of fluorescence intensity in wholemount imaging of oblique telophase cells (as performed for pMLC2, phalloidin, α18 and α-E-catenin in Figure 3) was performed by drawing linescans in the en-face perspective around the entire cortex at a series of focal planes: apical plane of the apical daughter, 'endfoot' of the apical daughter, mid-cell of the basal daughter and mid-cell of an interphase basal neighboring cell. For each image, background levels were determined using mean fluorescent intensity for each channel in a neighboring cell nucleus. This background value was subtracted from the mean cortical intensity of the appropriate channel. Fluorescence intensity ratios were quantified using background subtracted mean fluorescence intensities.

Cell density and live cell division orientation. Local cell density was calculated by counting the number of neighboring cells and dividing this number by their area. Area was measured in a single z-plane determined to be the cell center by orthogonal slices. Regions where the tissue sloped at an extreme angle were excluded due to inability to capture cell centers for all neighbors.

Differentiation/stratification analyses

Request a detailed protocol

E14.5 differentiation was quantified by imaging ~10 regions of the backskin in sagittal sections stained with β4 integrin, K10 and H2B-RFP. For each region, the number of basal and suprabasal cells were counted and the length of the region measured by the length of the underlying basement membrane (to account for tissue wrinkling/curvature). To quantify cell density, cell counts were divided by the length of each region in microns. At later stages (E16.5 or E18.5) we quantified K10 thickness by imaging ~10 regions of the backskin in sagittal sections stained with β4 integrin, K10 and H2B-RFP. Using the K10 channel, a thresholded, binary mask was created and filled, then used to measure the area above threshold. This thresholded area was then divided by the length of the underlying basement membrane (measured by β 4 integrin stain). n values for these analyses are representative of the number of regions imaged.

Mitotic index

Request a detailed protocol

Mitotic Index was quantified at E14.5 and E16.5 by imaging the entire length of sagittally-sectioned backskin stained for β4 and pHH3. pHH3+ basal cells were counted and the length of the entire backskin was measured by length of the underlying basement membrane. n values are indicative of the number of individual embryos analyzed.

All statistical analyses and graphs were generated using GraphPad Prism 8 and Origin 2015 (OriginLab). Error bars represent standard error of the mean (s.e.m.) unless otherwise noted. Statistical tests of significance were determined by Mann-Whitney U-test (non-parametric) or student’s t-test (parametric) depending on whether the data fit a standard distribution (determined by pass/fail of majority of the following: Anderson-Darling, D’Agostino & Pearson, Shapiro-Wilk, and Kolmogorov-Smirnov tests). Cumulative frequency distributions were evaluated for significant differences by Kolmogorov-Smirnov test. χ2 tests were utilized to evaluate expected (control) against experimental distributions of categorical values (e.g. LGN apical/absent/other distributions). All box-and-whisker plots are displayed as Tukey plots where the box represents the interquartile range (IQR, 25th-75th percentiles) and the horizontal line represents the median. Whiskers represent 1.5x IQR unless this is greater than the min or max value. Figures were assembled using Adobe Photoshop and Illustrator CC 2017.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
    Sonic hedgehog regulates growth and morphogenesis of the tooth
    1. HR Dassule
    2. P Lewis
    3. M Bei
    4. R Maas
    5. AP McMahon
    (2000)
    Development 127:4775–4785.
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62
  63. 63
  64. 64
  65. 65
  66. 66
  67. 67
  68. 68
  69. 69
  70. 70
  71. 71
  72. 72
  73. 73
  74. 74
  75. 75
  76. 76
  77. 77
  78. 78
  79. 79
  80. 80
  81. 81
  82. 82
  83. 83
  84. 84
  85. 85
  86. 86
  87. 87
  88. 88
  89. 89
  90. 90

Decision letter

  1. Valerie Horsley
    Reviewing Editor; Yale University, United States
  2. Anna Akhmanova
    Senior Editor; Utrecht University, Netherlands
  3. Alpha Yap
    Reviewer; University of Queensland, Australia
  4. Terry Lechler
    Reviewer; Duke University, United States
  5. Carien M Niessen
    Reviewer; University of Cologne, Germany

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

This manuscript reports and characterizes an interesting mechanism to regulate the orientation of cell division within epithelial tissues. Surprisingly, the authors find that additional mechanisms correct the orientation of a dividing cell in the later transition to telophase rather than the more well-studied anaphase. Interestingly, the authors implicate adherens junction proteins in this phase of spindle orientation. This work will be of interest to scientists in multiple fields including cellular adhesion, stem cell biology, and epithelial biology.

Decision letter after peer review:

Thank you for submitting your article "Telophase correction refines division orientation in stratified epithelia" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Anna Akhmanova as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Alpha Yap (Reviewer #1); Carien M Niessen (Reviewer #3).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

This manuscript reports and characterizes an interesting mechanism to regulate the orientation of cell division. Whereas much of this field has concentrated on how spindle orientation may be defined at the onset of anaphase, the authors show that there are additional mechanisms that correct orientation of the dividing cell in the later transition to telophase. This is prompted by the observation that cells are found with intermediate orientations at anaphase, which is corrected by telophase. The authors show that such correction involves two mechanisms: 1) A role for LGN in "perpendicular" correction, that extends its current role in anaphase specification; and 2) A role for the AJ cytoskeleton, that functionally involves α-catenin, vinculin and afadin. The latter is interesting because it appears to be dominant: although LGN-depleted cells can correct to planar orientation (perhaps because of this junctional input), correction appears to be fundamentally compromised (division planes are close-to randomized) when these AJ-actin linkers are depleted.

Characterization of this phenomenon of orientation correction is a valuable contribution to the field. This value is enhanced because it has been i) performed in a complex tissue model; and ii) in a stratified epithelium where decisions of perpendicular vs. planar division have consequences for cell fate (as the authors pursue in Figure 7).

Essential revisions:

1) The authors should provide more data regarding the physiological importance of telophase rescue. Is the cortex of either the dividing cell or its immediate neighbours altered in e.g. afadin and vinculin depletion? The authors have studied this in an in vitro model of junction formation, but one might suspect that the orientation correction mechanism involves generating some anisotropy in the cortex, e.g. concentration of Myosin II. It is difficult to discern the cytokinetic furrow in most videos and to clearly tell when abscission occurs. This is especially true for planar rescues, which could as likely be to cell respreading as cells exit mitosis. Along these lines, it would be nice to include some of the H2B-RFP videos in addition to the cell membrane videos that are included in the supplementary data section.

2) There is very little analysis of phenotypes resulting from loss of telophase rescue. Thus, the physiological significance of this process is questionable. Is the same precocious differentiation seen in vinculin mutant and acatenin mutants as well as afadin. Do they have later phenotypes that are consistent with spindle orientation functions? In addition, some of the work linking these proteins to junctional tension/organization comes only from cultured keratinocytes though the mutants are in hand. It would be ideal to see adherens junction/actin/tension markers in intact skin.

3) This work seems somewhat at odds with other reports in the field, including those from the PI. For example, LGN knockdown cells were reported to have almost entirely planar divisions, yet here they are reported to be quite random until telophase. Were those previous spindle orientation measurements all performed in telophase? And what about divisions orientation quantitations from multiple other groups that have reported more clear spindle orientations during anaphase?

4) Ultimately, one would like to know how to "order" the contributions of α-catenin, vinculin and afadin in this phenomenon. For example, at first glance it is slightly surprising that vinculin should decrease apparent tension across α-catenin, when the recruitment of vinculin to α-catenin is thought to be promoted by application of tension to the latter. This might reflect a contribution of vinculin to forces applied on integrin adhesion (which are secondarily transmitted to junctions) or simply the difficulty of parsing pathways that may not be linear. I mention this purely as a comment out of interest.

5) Revised discussion of the role vinculin in integrins. The authors don't directly address whether vinculin function at integrin or cadherin adhesions are required for telophase rescue. This should be directly stated.

The authors don't comment on whether vinculin knockdown (and other mutants) have a significant effect on perpendicular spindle telophase rescue. It appears from the data that these might affect perpendicular as well as parallel spindles. If so, this should be discussed.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Telophase correction refines division orientation in stratified epithelia" for further consideration by eLife. Your revised article has been evaluated by Anna Akhmanova as the Senior Editor, a Reviewing Editor and three reviewers.

The manuscript has been improved but there are some remaining issues that need to be addressed before acceptance, as outlined below:

The reviewers thought that the conclusions on tension regulation are too strong and should be tuned down in the text and legend titles. Once these revisions are complete, the manuscript will be appropriate for publication in eLife.

Reviewer #1:

I think the authors have reasonably responded to the questions raised in my earlier review. This is now a near-gigantic manuscript that reports an interesting phenomenon. Given its size and depth one can also quibble about details. But I think that the field would be better served by having the story out for consideration and replication.

Reviewer #2:

Substantial changes have been made, both experimentally and in writing, that strengthens the manuscript. I think it is largely suitable for publication.

I really wish that they would have included analysis of tissue thickness/premature differentiation of vinculin mutants, which would have strengthened the results (this data should be in hand).

I do have some reservations in inconsistency with the data with much other published work including the α-catenin knockdown data and the data on delamination at early stages of stratification. The Devenport lab does not see this with live imaging (and neither do we), although the Wickstrom group has. Since they performed imaging at E14.5, it would be great if they could state whether the rate of delamination they see in their videos validates their lineage tracing.

Reviewer #3:

This paper is improved and has addressed some of my major concerns. The lineage tracing data have helped to show that there is a likely direct consequence in terms of cell density in the different layers, even though if I understand Figure 7G and H correctly, this seems to have no direct effect on epidermal thickness or differentiation (time point of analysis is not mentioned though in the legends of these panels).

I do still feel that the authors conclusions on tension regulation are too strong as it is all based on staining (Figures 3 and 4) and should be tuned down in the text and legend titles as no experiment directly addresses the role of tension or hyper contractility. For example, the phosphomyosin staining is suggestive of increased contractility, but there is no direct evidence for this (laser ablation, etc.) and in addition the resolution in the images does not really allow one to make this very clear statement as in the legend title that a hyper contractile basal endfoot directs planar telophase correction. I realise that such experiments, especially in vivo are not trivial, but even in none of their in vitro experiments the authors directly test tension, but it is all based on staining. The authors should not misunderstand me, but the strength is the telophase correction pathway and the role of junctions in this pathway that is independent of LGN, but whether these are direct consequences of tension is beyond this manuscript and can be suggested and discussed but requires quite a bit more to show that convincingly.

Having said that, and if the authors are willing to tune those conclusions down, I think this is a very nice contribution to eLife.

https://doi.org/10.7554/eLife.49249.sa1

Author response

Essential revisions:

1) The authors should provide more data regarding the physiological importance of telophase rescue. Is the cortex of either the dividing cell or its immediate neighbours altered in e.g. afadin and vinculin depletion? The authors have studied this in an in vitro model of junction formation, but one might suspect that the orientation correction mechanism involves generating some anisotropy in the cortex, e.g. concentration of Myosin II. It is difficult to discern the cytokinetic furrow in most videos and to clearly tell when abscission occurs. This is especially true for planar rescues, which could as likely be to cell respreading as cells exit mitosis. Along these lines, it would be nice to include some of the H2B-RFP videos in addition to the cell membrane videos that are included in the supplementary data section.

We have addressed the reviewers’ comment in several ways:

We have analyzed anisotropy of phosphorylated (Ser19) myosin light chain 2 (pMLC2) within oblique telophase cells displaying a characteristic basal endfoot as observed in planar correction via live imaging. These analyses were performed by whole mount staining E16.5 epidermis with phalloidin and pMLC2 and quantifying fluorescence intensity in rare, obliquely oriented telophase divisions (Figure 3D). These measurements demonstrated a greater accumulation of pMLC2 in the basal endfoot than compared to the apical cortex of the “correcting” daughter cell, suggesting anisotropic contractility during telophase correction.

Further supporting this conclusion, we stained E16.5 epidermal whole mounts with both the a18 and a polyclonal α-E-catenin antibody to estimate the relative abundance of the open, “tensile” conformation of α-E-catenin in the same context. This quantification (performed similarly to the methodology for pMLC2) demonstrated that the a18 epitope stained with greater intensity in the basal endfoot compared to the apical cortex of the “correcting” daughter cell, while total a-Ecatenin levels were unchanged between these two regions (Figure 3F; Figure 3—figure supplement 1E). Furthermore, the a18: α-E-catenin fluorescence intensity ratio was highest in the basal endfoot when compared to other mitotic scenarios (Figure 3G). This suggests that the increased contractility in the endfoot changes the conformation of α-E-catenin, thereby justifying our pursuit of these molecules as candidate regulators of telophase correction.

Given the rarity of these oblique, anaphase/early telophase divisions in fixed tissue we are unable to make clear conclusions regarding the dynamics of these molecules through mitotic progression. Ultimately, elucidating these mechanisms will likely require live imaging of fluorescently labeled transgenes for α-E-catenin, afadin, vinculin, myosin, and/or actin. We attempted such an experiment utilizing a miRFP-670 conjugated LifeAct construct packaged in high-titer lentivirus, but were unable to complete imaging due to difficulty in either A) achieving a useable infection rate or B) inability to detect dim fluorescent signal from the construct on the available microscopes. We are unaware of anyone successfully performing live-imaging of lentivirally-expressed fluorescent transgenes, other than our own success with the extremely bright H2B-RFP reporter. We predict that many of these transgenes will be relatively lowly expressed, making fluorescence detection difficult in thick, stratified epidermis (E16.5). While this presents an exciting avenue of future research, approaching these experiments will require substantial technical optimization in terms of viral packaging, expression, and image collection which we consider beyond the scope of the current manuscript.

2) There is very little analysis of phenotypes resulting from loss of telophase rescue. Thus, the physiological significance of this process is questionable. Is the same precocious differentiation seen in vinculin mutant and acatenin mutants as well as afadin. Do they have later phenotypes that are consistent with spindle orientation functions? In addition, some of the work linking these proteins to junctional tension/organization comes only from cultured keratinocytes though the mutants are in hand. It would be ideal to see adherens junction/actin/tension markers in intact skin.

We have addressed the reviewers’ comment in several ways:

We have analyzed the precocious differentiation phenotype at E14.5 for Ctnna1 knockdown samples (Ctnna1912), which present with a nearly identical phenotype to the Afdn knockdown, suggesting the precocious differentiation observed in these mutants is more widely attributable to those mutants which fail to undergo telophase correction.

We have similarly analyzed the effects on differentiation at E16.5 for the following genotypes: Afdn2711, Afdnfl/fl + lenti-Cre and Ctnna1912. all of which display sustained hyperstratification. Similarly, the Afdn2711, Afdnfl/fl + lenti-Cre increased thickness further persists through E18.5. These data suggest that failed telophase correction increases differentiation throughout epidermal stratification.

We have also investigated the potential impact of increased proliferation on the observed differentiation phenotypes. While previous reports from α-E-catenin knockout epidermis confirmed a hyperproliferative phenotype, the same study (Berojna and Fuchs, 2010; cited numerous times) concluded that Ctnna1 knockout clones were at a net growth disadvantage due to increased apoptosis. Similarly, these analyses were restricted to BrdU incorporation assays, which, while frequently correlative with rates of cell division, do not directly address rates of mitosis and could possibly be due to lengthened S-phase. Here, we have quantified mitotic index by counting pHH3+ cells per mm of backskin and shown at E14.5 and E16.5, that neither loss of Afdn nor Ctnna1 results in a significant increase in mitotic events.

Lastly, we have performed short-term genetic lineage tracing experiments for Afdn knockdown and WT littermates (Figure 8I-K; Figure 8—figure supplement 1D-G) to examine clonal growth during stratification (E14.5-E17.5). Using this assay we were able to conclude the following:

– Loss of afadin results in clones with a greater number of differentiated suprabasal cells with no significant change in overall clone size, suggesting a shift towards asymmetric cell divisions, likely due to failure of oblique divisions to undergo planar correction.

– Delamination – an alternative differentiation pathway to asymmetric cell divisions – is largely unaffected by Afdn loss. In our clonal analyses, we assumed that suprabasal clones lacking any basal cells were likely the result of differentiation by delamination (without cell division). Given this assumption, loss of afadin did not significantly alter the rate of delamination, further suggesting that the increased stratification in these mutants is due specifically to errors in telophase correction.

– As WT and afadin knockdown clones had a similar number of total labeled cells (Figure 8—figure supplement 1E) we were able to reaffirm that loss of afadin had no impact on net growth.

Ultimately, these data further support the conclusion that the precocious and sustained hyperstratification observed in both mutants is most likely due to failed telophase correction. These data form the centerpiece of a newly added Figure 8, which focuses on characterizating the consequences of failed telophase correction. Pertinent discussion of these results has been added to a subheader in the Results section titled “Telophase correction impacts cell fate decisions”.

3) This work seems somewhat at odds with other reports in the field, including those from the PI. For example, LGN knockdown cells were reported to have almost entirely planar divisions, yet here they are reported to be quite random until telophase. Were those previous spindle orientation measurements all performed in telophase?

We thank the reviewers for pointing out the apparent contradiction between this study and previous results. We realize we did not contextualize our findings as clearly as we could have, so we have expanded on this in the first paragraph of the Results section:

“Our previous studies reported a bimodal distribution of division angles at late stages of mitosis and randomized division angles during metaphase (Williams et al., 2011; Williams et al., 2014), while other groups have reported that spindle rotation occurs during prometaphase and is fixed to a bimodal distribution by late metaphase/early anaphase (Poulson and Lechler, 2010; Seldin et al., 2016). While these studies agree that spindle rotation occurs, they come to different conclusions about when and how the spindle axis becomes fixed to a bimodal distribution.”

We further explain the rationale for our chosen method of using Survivin to label late-stage mitotic cells in the subsection “Randomized division orientation persists into anaphase”.

To improve the rigor and reproducibility of our studies of metaphase/anaphase/telophase spindle orientation in fixed tissue, we included three separate commonly-used strains of mice (CD1, C57 and 129), and separately analyze each strain (Figure 1—figure supplement 1A). Since each strain showed similar patterns, we rule out potential strain differences as an explanation for any reported differences in anaphase division orientation.

And what about divisions orientation quantitations from multiple other groups that have reported more clear spindle orientations during anaphase?

While there are examples of spindle or division orientation measurements made at metaphase or anaphase in the epidermis, each varies slightly in method from ours reported here. These are discussed in detail below:

Lechler and Fuchs, 2005 (cited numerous times): In Figure 1A-E the authors present data which first reported the presence of symmetric-planar and asymmetric-perpendicular divisions in the embryonic epidermis. These experiments utilized either DAPI and tubulin (Figure 2A,B, E) or DAPI with a Centrin-GFP allele (Figure 2C-E) to measure division angles. However, these data to not differentiate between anaphase and telophase.

Poulson et al., 2010 (cited numerous times): In Figure 2C-E the authors present data measuring spindle angles in vivo using Centrin-GFP or NuMA-GFP alleles. At prometaphase (Figure 2C-D) they demonstrate nearly random spindle orientation but show a dramatic shift towards bipolar angles at metaphase (Figure 2E). However, it is important to note that the measurements were made using an overexpressed spindle orientation effector, NuMA-GFP which possibly impacts division orientation behavior.

Seldin et al., 2016 (now cited): In Figure 3J the authors present data measuring spindle angles using IF staining for pHH3 to ID anaphase cells in E17.5 interfollicular epidermis. While the authors of this study state measurements were restricted specifically to anaphase, they do not present example images, making the discrepancy difficult to explore. The described methods suggest that anaphase cells were identified by pHH3 staining, which may persist into telophase (data delineating pHH3 antibody specificity through distinct stages of mitosis in epidermal progenitors does not exist, and in our hands has proven to be variable between distinct antibody). These measurements were also performed on thin sagittal sections, limiting the authors ability to observe both planar and oblique division angles, which may frequently occur out of the section plane. Lastly, these measurements were taken at E17.5, one day later than the data presented in this manuscript, which may alter the timing of division plane commitment.

We are unaware of other instances of metaphase or anaphase measurements of spindle/division orientation in the embryonic epidermis and would welcome more specific citations if we have missed a relevant data set. However, the above listed studies do not present data specifically measuring division orientation at anaphase in E16.5 or E14.5 interfollicular epidermis. The data evaluating metaphase/anaphase division angles were either collected in a mutant context overexpressing a key spindle orientation gene, or at a different timepoint (E17.5) using different methods of quantitation. Similarly, these data likely underestimate the proportion of oblique and planar divisions due to the fact that these measurements were made in sagittal sections, which impact the ability to observe divisions which occur out of the plane of the section.

4) Ultimately, one would like to know how to "order" the contributions of α-catenin, vinculin and afadin in this phenomenon. For example, at first glance it is slightly surprising that vinculin should decrease apparent tension across α-catenin, when the recruitment of vinculin to α-catenin is thought to be promoted by application of tension to the latter. This might reflect a contribution of vinculin to forces applied on integrin adhesion (which are secondarily transmitted to junctions) or simply the difficulty of parsing pathways that may not be linear. I mention this purely as a comment out of interest.

The reviewers are correct to highlight that initial tension across α-E-catenin (inducing the “open” conformation) would likely be required for vinculin and afadin binding (based on the findings of Buckley, Nelson and Dunn, 2014; Science as well as Pokutta and Weiss, 2002). However, this model is not exclusive to the possibility that vinculin binding may stabilize the open conformation of α-E-catenin, which would result in a larger proportion of α-E-catenin molecules receiving the a18 label in a fixed stain. The same may be true for afadin as well. Our data suggest that these two proteins (vinculin and afadin) cooperatively promote the stabilization of α-E-catenin’s open, “tensile” conformation, which in turn promotes the formation of the afadin/α-E-catenin and vinculin/α-E-catenin interactions. One interesting observation that we make is that it is not so much that a18 levels decrease, but rather that total α-E-catenin levels increase upon Vcl or Afdn loss (Figure 4—figure supplement 1A; Figure 1—figure supplement 2G, H). This suggests there may be compensatory upregulation of α-E-catenin (and perhaps other AJ components) to counter the weakened AJs. This is discussed in the subsection “Telophase corrective basal contacts are hyper-contractile”.

We agree that our data are far from conclusive, but should note that there is a decade of literature addressing mechanistically how vinculin binding regulates the tension sensing capabilities of α-E-catenin, and very little work has been done on afadin. Although we lack the capabilities to tease their “order” of incorporation into the complex apart, we have analyzed the spindle orientation phenotype in double mutants of Vcl and Afdn, which resemble single mutants: “Examination of division orientation in single and double mutants revealed that vinculin loss did not exacerbate the Afdn cKO phenotype, suggesting that these proteins do not act additively in the context of division orientation (Figure 5I).” Our in vitro nascent AJ formation experiments lack the temporal resolution to identify a sequential recruitment into AJs, and we believe the biochemical analyses required for fully elucidate the underlying mechanisms of these binding events are beyond the scope of this current project. We would be happy to discuss the implications of our data regarding a molecular model of vinculin/afadin binding to α-E-catenin but feel any discussion beyond what is already included in the text would be highly speculative.

5) Revised discussion of the role vinculin in integrins. The authors don't directly address whether vinculin function at integrin or cadherin adhesions are required for telophase rescue. This should be directly stated.

The authors don't comment on whether vinculin knockdown (and other mutants) have a significant effect on perpendicular spindle telophase rescue. It appears from the data that these might affect perpendicular as well as parallel spindles. If so, this should be discussed.

We have addressed the reviewers’ comments in several ways.

We have added relevant discussion of vinculin in integrin based adhesions to the text. At the end of the first paragraph in the Discussion we have added the following statement: “Importantly, while our data support a model wherein vinculin regulates dynamic assembly of AJ, we cannot exclude the possibility that vinculin may play similar roles in cell-matrix integrin adhesions, which may also impact telophase correction.” We believe this statement is justified due to the fact that loss of α-E-catenin or afadin, neither of which have been implicated in cell-matrix adhesion, produce errors in telophase correction. However, the reviewers are correct to point out that we cannot completely exclude the possibility that vinculin serves roles at both cell-cell and cell-matrix adhesions, due in part to the difficulties in adequately labeling/staining vinculin in epidermal basal cells.

We have added specific comments highlighting that vinculin, α-E-catenin, and afadin loss results in errors to telophase correction in both planar and perpendicular directions.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

The reviewers thought that the conclusions on tension regulation are too strong and should be tuned down in the text and legend titles. Once these revisions are complete, the manuscript will be appropriate for publication in eLife.

We have made efforts throughout the text to temper our language with respect to the conclusions we draw about tension regulation. See specific examples in the response to reviewer #3, below.

Reviewer #2:

[…]

I really wish that they would have included analysis of tissue thickness/premature differentiation of vinculin mutants, which would have strengthened the results (this data should be in hand).

While we agree that the addition of analysis of Vcl mutants would have made a nice addition to Figure 8, unfortunately these were not data that we had “in hand.” It would have been necessary for us to generate additional Vcl mutants at E14.5, which would have delayed resubmission by at least another month. We did perform a cursory examination of E16.5 Vcl mutants and did not observe obvious differentiation defects. However, these embryos were more mosaic (thus containing less mutant tissue) than either of the other mutants. It should also be noted that even in Ctnna1 and Afdn mutants, the differentiation defect is more subtle at this age than at E14.5.

I do have some reservations in inconsistency with the data with much other published work including the α-catenin knockdown data and the data on delamination at early stages of stratification. The Devenport lab does not see this with live imaging (and neither do we), although the Wickstrom group has. Since they performed imaging at E14.5, it would be great if they could state whether the rate of delamination they see in their videos validates their lineage tracing.

I believe the only inconsistency between our findings and other published work regarding α-E-catenin is related to the expression of LGN in Ctnna1 mutants (Lechler and Fuchs, 2005). It is possible that this could be due to different antibodies that were used in these studies, the fact that different models of Ctnna1 loss were used, or potentially different ages that were examined. For example, we focused on E16.5 while it is not stated which age was examined in the Lechler & Fuchs study (Figure 4D).

Regarding delamination, in order to clarify this point, we have edited the Results section: “While we did not observe clear delamination events in our 3-6 hour live imaging experiments, genetic lineage tracing revealed that a similar and significant fraction of clones in both WT and Afdn2711 epidermis (48% vs. 40%) arose from delamination (Figure 8—figure supplement 1I).” However, we have not specifically searched for delamination events in our E14.5 videos since our focus was on mitotic cells, thus we do not feel comfortable concluding whether or not they occur, or at what frequency. To our knowledge, as acknowledged by this reviewer, the only lab that has reported observing delamination by live imaging is the Wickstrom lab (Miroshnikova et al., 2018), which used LifeAct-GFP in E15.5 embryos. It is possible this discrepancy between her lab and the others mentioned could be due to differences in imaging preparations, transgenic mouse lines, and/or age. However, we agree with the

Wickstrom lab with respect to the idea that delamination does occur, particularly in early stratification. Our lab has now utilized two different lineage tracing approaches – lentiviral Cre-RFP (Williams et al., 2014) and Krt14CreER (this study) – which have identified a significant and consistent proportion of clones that arise from apparent delamination events.

Reviewer #3:

This paper is improved and has addressed some of my major concerns. The lineage tracing data have helped to show that there is a likely direct consequence in terms of cell density in the different layers, even though if I understand Figure 7G and H correctly, this seems to have no direct effect on epidermal thickness or differentiation (time point of analysis is not mentioned though in the legends of these panels).

We applaud the reviewer for recognizing that the Krt14Cre; Afdnfl/flcKOs do not show increased stratification compared to Afdnfl/flWT controls (Figure 7G, H), which is at apparent odds with our later conclusion that Afdn2711 and lentiviral Cre-RFP Afdnfl/flcKOs show hyper-stratification (Figure 8—figure supplement 1D, E).However, we think this is most likely due to differences in the timing of Afdn targeting in these different lines. We have previously shown that lentiviral Cre (targeted to the amniotic fluid at E9.5) acts 2-3 days earlier than Krt14Cre (Beronja et al., 2010; Supplementary Figure 4). We have added a sentence to clarify this point in the Results section: “Moreover, in contrast to lentiviral-delivered Afdn2711, Krt14Cre-mediated deletion of Afdn did not cause obvious differentiation defects (compare Figure 7H to Figure 8—figure supplement 1E), likely because Afdn deletion occurs later with Krt14Cre (Beronja et al., 2010).”

I do still feel that the authors conclusions on tension regulation are too strong as it is all based on staining (Figures 3 and 4) and should be tuned down in the text and legend titles as no experiment directly addresses the role of tension or hyper contractility. For example, the phosphomyosin staining is suggestive of increased contractility, but there is no direct evidence for this (laser ablation, etc.) and in addition the resolution in the images does not really allow one to make this very clear statement as in the legend title that a hyper contractile basal endfoot directs planar telophase correction. I realise that such experiments, especially in vivo are not trivial, but even in none of their in vitro experiments the authors directly test tension, but it is all based on staining. The authors should not misunderstand me, but the strength is the telophase correction pathway and the role of junctions in this pathway that is independent of LGN, but whether these are direct consequences of tension is beyond this manuscript and can be suggested and discussed but requires quite a bit more to show that convincingly.

Having said that, and if the authors are willing to tune those conclusions down, I think this is a very nice contribution to eLife.

We have tempered our language regarding tension regulation throughout the text. Some examples are:

– Results header: “Telophase corrective basal contacts are hypercontractile” changed to “Telophase corrective basal contacts display hallmarks of elevated actomyosin contractility”

– Results: “This anisotropy suggests that the basal endfoot is hypercontractile during telophase correction and that this contractility may function to pull the apical daughter back into the basal layer.” changed to “This anisotropy suggests that the basal endfoot may be enriched in contractile actomyosin, which we speculate may serve the function of pulling the apical daughter back into the basal layer.”

– Results: “In agreement with the observed hypercontractility in the basal endfoot of oblique telophase cells…” changed to “In agreement with the observed increase in pMLC2 in the basal endfoot of oblique telophase cells…”

– Results: “These data suggest that increased actomyosin contractility and AJ-tension may play a significant role in planar directed telophase correction.” changed to “These data suggest that increased actomyosin contractility and associated conformational changes to α-E-catenin could play a role in planar directed telophase correction.”

– Results: “These data demonstrate that α-E-catenin and vinculin are required for the proper maturation of tensile AJs.” changed to “These data demonstrate that α-E-catenin and vinculin are required for the proper maturation of AJs.”

– Results: “Collectively, these data suggest that afadin is a novel regulator of AJ tension sensitivity by affecting α-E-catenin conformation and vinculin recruitment.” changed to “Collectively, these data suggest that afadin is a novel regulator of AJ maturation by affecting α-E-catenin conformation and vinculin recruitment.”

– Discussion: “…where the direction of correction is dependent on contact with the basement membrane via a hypercontractile basal endfoot.” changed to “…where the direction of correction is dependent on contact with the basement membrane via a basal endfoot.”

– Figure 3 Legend title: “Maintenance of basal contact through a tensile endfoot correlates with planar-directed telophase correction” changed to “Maintenance of basal contact correlates with planar-directed telophase correction”

– Figure 3—figure supplement 1 Legend title: “A hypercontractile basal endfoot directs planar telophase correction in WT and LGN knockdown samples” changed to “A basal endfoot mediates planar telophase correction”

– Figure 4 Legend title: “Vinculin, α-E-catenin and afadin regulate tension and AJ linkage to the actin cytoskeleton” changed to “Vinculin and afadin regulate α-E-catenin conformation and AJ linkage to the actin cytoskeleton”

https://doi.org/10.7554/eLife.49249.sa2

Article and author information

Author details

  1. Kendall J Lough

    1. Department of Pathology and Laboratory Medicine, Lineberger Comprehensive Cancer Center, The University of North Carolina, Chapel Hill, United States
    2. Department of Biology, Lineberger Comprehensive Cancer Centre, The University of North Carolina, Chapel Hill, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-9663-6983
  2. Kevin M Byrd

    1. Department of Pathology and Laboratory Medicine, Lineberger Comprehensive Cancer Center, The University of North Carolina, Chapel Hill, United States
    2. Department of Biology, Lineberger Comprehensive Cancer Centre, The University of North Carolina, Chapel Hill, United States
    3. Department of Oral & Craniofacial Health Sciences, The University of North Carolina School of Dentistry, Chapel Hill, United States
    Contribution
    Conceptualization, Investigation, Methodology
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-5565-0524
  3. Carlos P Descovich

    1. Department of Pathology and Laboratory Medicine, Lineberger Comprehensive Cancer Center, The University of North Carolina, Chapel Hill, United States
    2. Department of Biology, Lineberger Comprehensive Cancer Centre, The University of North Carolina, Chapel Hill, United States
    Contribution
    Data curation, Formal analysis, Methodology
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-6366-5195
  4. Danielle C Spitzer

    1. Department of Pathology and Laboratory Medicine, Lineberger Comprehensive Cancer Center, The University of North Carolina, Chapel Hill, United States
    2. Department of Biology, Lineberger Comprehensive Cancer Centre, The University of North Carolina, Chapel Hill, United States
    Present address
    Molecular and Cellular Biology Program, University of California, Berkeley, United States
    Contribution
    Data curation, Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4827-1857
  5. Abby J Bergman

    1. Department of Pathology and Laboratory Medicine, Lineberger Comprehensive Cancer Center, The University of North Carolina, Chapel Hill, United States
    2. Department of Biology, Lineberger Comprehensive Cancer Centre, The University of North Carolina, Chapel Hill, United States
    Contribution
    Data curation, Investigation
    Competing interests
    No competing interests declared
  6. Gerard MJ Beaudoin III

    1. Department of Biochemistry & Biophysics, University of California, San Francisco, San Francisco, United States
    2. Department of Physiology, University of California, San Francisco, San Francisco, United States
    Present address
    Department of Biology, Trinity University, San Antonio, United States
    Contribution
    Resources
    Competing interests
    No competing interests declared
  7. Louis F Reichardt

    1. Department of Biochemistry & Biophysics, University of California, San Francisco, San Francisco, United States
    2. Department of Physiology, University of California, San Francisco, San Francisco, United States
    Contribution
    Resources
    Competing interests
    No competing interests declared
  8. Scott E Williams

    1. Department of Pathology and Laboratory Medicine, Lineberger Comprehensive Cancer Center, The University of North Carolina, Chapel Hill, United States
    2. Department of Biology, Lineberger Comprehensive Cancer Centre, The University of North Carolina, Chapel Hill, United States
    Contribution
    Conceptualization, Resources, Supervision, Investigation, Methodology, Project administration
    For correspondence
    scott_williams@med.unc.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-9975-7334

Funding

National Institute of Dental and Craniofacial Research (Predoctoral Fellowship F31 DE026956)

  • Kendall J Lough

National Institute of Dental and Craniofacial Research (Career Development Award K08 DE026537)

  • Kevin M Byrd

Sidney Kimmel Foundation for Cancer Research (Scholar Award SKF-15-165)

  • Scott E Williams

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank members of the Williams and Peifer Labs for their critical feedback. We thank Dr. Danelle Devenport (Princeton) for graciously sharing her live imaging protocol. We thank Dr. Akira Nagafuchi (Nara Medical University) for sharing the α18 antibody. We thank Dr. Brent Hoffman and Evan Gates (Duke University) for their effort and feedback regarding E-cadherin mediated junctional tension. We thank Dr. Keith Burridge (UNC) for sharing the vinculin (Rb) polyclonal antibody. We thank Kendra Niederkorn for design input into the model in Figure 9. KJL was supported by an NIH Ruth L Kirschstein Predoctoral National Research Service Award (F31 DE026956). KMB is supported by an NIH/NIDCR K08 Mentored Clinical Scientist Research Career Development Award (DE026537). SW was supported by a Sidney Kimmel Scholar Award (SKF-15-165).

Ethics

Animal experimentation: Mice were housed in an AAALAC-accredited (#329; June 2017), USDA registered (55-R-0004), NIH welfare-assured (D16-00256 (A3410-01)) animal facility. All procedures were performed under IACUC-approved animal protocols (19-155).

Senior Editor

  1. Anna Akhmanova, Utrecht University, Netherlands

Reviewing Editor

  1. Valerie Horsley, Yale University, United States

Reviewers

  1. Alpha Yap, University of Queensland, Australia
  2. Terry Lechler, Duke University, United States
  3. Carien M Niessen, University of Cologne, Germany

Publication history

  1. Received: June 11, 2019
  2. Accepted: December 12, 2019
  3. Accepted Manuscript published: December 13, 2019 (version 1)
  4. Version of Record published: January 14, 2020 (version 2)

Copyright

© 2019, Lough et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,021
    Page views
  • 213
    Downloads
  • 3
    Citations

Article citation count generated by polling the highest count across the following sources: PubMed Central, Crossref, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Biochemistry and Chemical Biology
    2. Cell Biology
    Mikel Garcia-Marcos et al.
    Tools and Resources Updated

    Heterotrimeric G-proteins are signal transducers involved in mediating the action of many natural extracellular stimuli and many therapeutic agents. Non-invasive approaches to manipulate the activity of G-proteins with high precision are crucial to understand their regulation in space and time. Here, we developed LOV2GIVe, an engineered modular protein that allows the activation of heterotrimeric G-proteins with blue light. This optogenetic construct relies on a versatile design that differs from tools previously developed for similar purposes, that is metazoan opsins, which are light-activated G-protein-coupled receptors (GPCRs). Instead, LOV2GIVe consists of the fusion of a G-protein activating peptide derived from a non-GPCR regulator of G-proteins to a small plant protein domain, such that light uncages the G-protein activating module. Targeting LOV2GIVe to cell membranes allowed for light-dependent activation of Gi proteins in different experimental systems. In summary, LOV2GIVe expands the armamentarium and versatility of tools available to manipulate heterotrimeric G-protein activity.

    1. Cell Biology
    2. Plant Biology
    Madlen Stephani et al.
    Research Article Updated

    Eukaryotes have evolved various quality control mechanisms to promote proteostasis in the endoplasmic reticulum (ER). Selective removal of certain ER domains via autophagy (termed as ER-phagy) has emerged as a major quality control mechanism. However, the degree to which ER-phagy is employed by other branches of ER-quality control remains largely elusive. Here, we identify a cytosolic protein, C53, that is specifically recruited to autophagosomes during ER-stress, in both plant and mammalian cells. C53 interacts with ATG8 via a distinct binding epitope, featuring a shuffled ATG8 interacting motif (sAIM). C53 senses proteotoxic stress in the ER lumen by forming a tripartite receptor complex with the ER-associated ufmylation ligase UFL1 and its membrane adaptor DDRGK1. The C53/UFL1/DDRGK1 receptor complex is activated by stalled ribosomes and induces the degradation of internal or passenger proteins in the ER. Consistently, the C53 receptor complex and ufmylation mutants are highly susceptible to ER stress. Thus, C53 forms an ancient quality control pathway that bridges selective autophagy with ribosome-associated quality control in the ER.