1. Cell Biology
  2. Developmental Biology
Download icon

Cellular aspect ratio and cell division mechanics underlie the patterning of cell progeny in diverse mammalian epithelia

  1. Kara L McKinley
  2. Nico Stuurman
  3. Loic A Royer
  4. Christoph Schartner
  5. David Castillo-Azofeifa
  6. Markus Delling
  7. Ophir D Klein  Is a corresponding author
  8. Ronald D Vale  Is a corresponding author
  1. University of California, San Francisco, United States
  2. Chan Zuckerberg Biohub, United States
Research Article
  • Cited 0
  • Views 2,608
  • Annotations
Cite as: eLife 2018;7:e36739 doi: 10.7554/eLife.36739

Abstract

Cell division is essential to expand, shape, and replenish epithelia. In the adult small intestine, cells from a common progenitor intermix with other lineages, whereas cell progeny in many other epithelia form contiguous patches. The mechanisms that generate these distinct patterns of progeny are poorly understood. Using light sheet and confocal imaging of intestinal organoids, we show that lineages intersperse during cytokinesis, when elongated interphase cells insert between apically displaced daughters. Reducing the cellular aspect ratio to minimize the height difference between interphase and mitotic cells disrupts interspersion, producing contiguous patches. Cellular aspect ratio is similarly a key parameter for division-coupled interspersion in the early mouse embryo, suggesting that this physical mechanism for patterning progeny may pertain to many mammalian epithelia. Our results reveal that the process of cytokinesis in elongated mammalian epithelia allows lineages to intermix and that cellular aspect ratio is a critical modulator of the progeny pattern.

https://doi.org/10.7554/eLife.36739.001

eLife digest

The body has an impressive ability to renew itself by replacing old and damaged cells with new ones. This can happen rapidly; for example, the lining of the intestine renews itself approximately every five days. The lining contains many different cell types, which exchange important signals with their neighbors. This means that the new cells need to occupy similar positions to the ones they are replacing to keep the intestine working.

New cells form when existing cells double their contents and divide. In many tissues the resulting cells sit side-by-side. But when cells in the intestine divide, the new cells often separate, ending up on either side of a cell that did not divide.

To investigate how this happens, McKinley et al. used live microscopy techniques to watch in real time as new cells divide and position themselves in mouse intestinal organoids – miniature versions of organs that can be grown outside the body. This revealed that the shape of intestinal cells explains why the newly formed cells become separated. Intestinal cells are taller than they are wide, and divide near their top edge. This enables a neighboring cell to squeeze between the new cells as they divide.

Further experiments showed that tall cells in other mouse tissues also become separated after division. The process of new cells interspersing with their neighbors due to their height is therefore not unique to the intestine. It may also be common in other mammalian tissues. There is great potential for investigating this further because labs can now grow many types of organoids, representing different organs. Using live microscopy to examine them could reveal more about how various tissues grow.

https://doi.org/10.7554/eLife.36739.002

Introduction

Epithelia are sheets of polarized cells that function as barriers between compartments of multicellular organisms and between the organism and the external environment. In addition to providing a physical barrier, specialized epithelial cell types provide functions including sensation, absorption and secretion, and contribute to the identities of nearby cells through cell-cell signaling. Proper epithelial function requires that these diverse cell types are positioned appropriately within the tissue and that this distribution is maintained as new cells are added through cell division.

The adult mammalian small intestine is a prime example of an epithelium that contains many cell types and maintains a high degree of spatial organization during rapid turnover (Barker, 2014). In the small intestine, divisions of stem cells in the crypts of Lieberkühn replenish the stem cell pool and generate absorptive and secretory progenitor cells in the crypt, which in turn produce differentiated cells that carry out the absorptive and protective functions of the gut (Gracz and Magness, 2014). Throughout the epithelium, cells derived from a given progenitor intersperse with other cells (Carroll et al., 2017). In particular, lineage tracing in fixed tissues has established that cells derived from secretory progenitors intermix with cells derived from absorptive progenitors along the crypt and villus length (Yang et al., 2001). At the crypt base, stem cells are interspersed with Paneth cells (Farin et al., 2016). Interspersion of cell lineages plays important roles in determining local signaling environments required for intestinal homeostasis. For example, intestinal stem cells receive signals critical to their identity from neighboring Paneth cells (Sato et al., 2011). Indeed, direct contact between stem and Paneth cells supports stem cell maintenance (Farin et al., 2016). However, the molecular mechanisms that underlie the intermixing of lineages are poorly understood.

Here, we use light sheet and confocal imaging of live murine small intestinal organoids to define the mechanisms of cell interspersion. We find that rearrangements of the actin cytoskeleton displace mitotic cells along the apical-basal axis, such that cell division occurs at the apical surface. Interspersion arises when elongated interphase neighboring cells wedge between apically dividing daughters during cytokinesis. We find that the propensity to intersperse during division requires an elongated shape of cells in the epithelium; reducing the cellular aspect ratio (height: width) in organoids disrupts interspersion, resulting in outgrowth of lineage patches. Consistent with our data indicating that the physical parameters of the tissue are a critical determinant of interspersion during division, we demonstrate that the elongated epiblast/primitive ectoderm of post-implantation (E7.5) mouse embryos, but not the short visceral endoderm, also undergoes division-coupled cell interspersion. Thus, tissues of distinct developmental context from the adult small intestine exhibit similar mechanisms for patterning cellular progeny according to cellular dimensions. Together, our data indicate that cell shape differences between interphase and mitotic cells in elongated mammalian epithelia can allow a neighboring cell to insert between nascent daughter cells during cytokinesis and drive interspersion of cellular progeny.

Results

Cells of different lineages intersperse during cell division in the intestinal epithelium

To identify the basis for cell interspersion, we performed time-lapse imaging of adult murine small intestinal organoids (Kretzschmar and Clevers, 2016; Sato et al., 2009) by confocal and light sheet microscopy (single plane illumination microscopy - SPIM) (Wu et al., 2013) (Figure 1A). To visualize cell lineages, we first used organoids in which the cytoplasm of cells of the secretory lineage was labeled with RFP (Atoh1CreER; R26RFP). Strikingly, we observed that daughter cells separated from one another in approximately half of divisions (31/50 divisions, Figure 1A and Video 1; also see [Carroll et al., 2017]). We observed that Atoh1-expressing secretory daughters along the crypt length separated from one another, mixing with unlabeled cells (Figure 1A and Video 1). 3D SPIM data confirmed that cells were fully separated on their basal surface, although they maintained a minimal contact on the apical surface, creating a V-shaped geometry (Figure 1C, Figure 1—figure supplement 1E, Figure 1—video 1, 9/16 daughter pairs). When daughters did not separate during the division (Figure 1D, top panels, 7/16 daughter pairs), these daughters either became separated at later time points by division of a neighboring cell (Figure 1D, bottom panels and Figure 1—video 2), or remained as neighbors for the duration of imaging. These data indicate that separation of nascent daughter cells during cell division makes substantial contributions to the relative positioning of cell types within the intestinal epithelium.

Figure 1 with 3 supplements see all
Separation of dividing daughter cells during apical cytokinesis underlies intermingling of cell lineages.

(A) Left: Cartoon depicting organoid derivation. Right: Frames from time-lapse imaging of a dividing cell of the secretory lineage (red, Atoh1CreER; R26RFP) interspersing with non-secretory cells (green membranes). Arrowheads: dividing cell. Fraction of divisions in which labeled daughters separated is shown on the right panel. (B) Frames from 3D reconstructed SPIM of a secretory cell (red, Atoh1CreER; R26RFP) inserting in the cytokinetic furrow of a dividing stem cell (green, Lgr5DTR-GFP). Arrows: dividing cell. (C) Frames from 3D reconstructed SPIM of a dividing cell of the secretory lineage (Atoh1CreER; R26RFP). Arrowheads: dividing cell. (D) Frames from 3D reconstructed SPIM of a secretory cell (red) undergoing a division in which daughter cells do not separate during cytokinesis (top, white arrows indicate daughter cells). Subsequently, these daughter cells become separated by a dividing cell pushing between them (bottom, white arrows indicate daughter cells and yellow arrowhead indicates newly dividing cell inserting between the adjacent daughters). (E) Confocal images of crypts in which cells have been labeled with a stochastic multicolor reporter in vivo (Vil1CreER; R26Brainbow2.1) and the positions of progeny analyzed three days after induction of the reporter. Left: sagittal view from 50 µm sections. Right: transverse views from 20 µm sections. Arrowheads indicate interspersed progeny. Progeny can also remain adjacent, as in the organoids, indicated by asterisks. Scale bars, 10 µm.

https://doi.org/10.7554/eLife.36739.003
Video 1
Secretory cell separation during division.

Cells of the secretory lineage (red, Atoh1CreER; R26RFP) interspersed with non-secretory cells (green membranes) imaged by spinning disc confocal with 20X objective at 3 min time points.

https://doi.org/10.7554/eLife.36739.007

We next tested whether daughter cell separation was a common feature of cell lineages in the intestinal epithelium. Notch1-expressing cells (from Notch1CreERT2; R26RFP organoids), which comprise all non-secretory cells including stem cells and absorptive cells, also interspersed during division (Figure 1—figure supplement 1A). Finally, dividing stem cells (labeled with Lgr5DTR-GFP) at the crypt base also separated, with secretory (Paneth) cells (labeled with Atoh1CreER; R26RFP) inserting between them (Figure 1B, Figure 1—figure supplement 1B and Video 2). Altering cell fates, for example by inhibiting Notch signaling to cause an expansion of secretory cells, did not alter the frequency of this process (Figure 1—figure supplement 1C,D). Thus, cells intersperse during a subset of divisions in all cell lineages of the crypt epithelium.

Video 2
Stem cell separation during division by insertion of a secretory cell into the cytokinetic furrow.

Cell of the secretory lineage (red, Atoh1CreER; R26RFP) inserts into the furrow of a dividing stem cell (green, Lgr5DTR-GFP). Imaged by SPIM with 40X objectives at 2 min time points. Second clip isolates only the cell of the secretory lineage.

https://doi.org/10.7554/eLife.36739.008

We next sought to determine whether the interspersion of cellular progeny observed in organoids also occurred in the intestine in vivo. To this end, we labeled a subset of cells in the intestines of adult mice with different fluorophores by induction of the stochastic multicolor reporter allele, R26Brainbow2.1 (Vil1CreERT2; R26Brainbow2.1). After three days of Cre induction, which is sufficient for most crypt epithelial cells to divide at least once (Snippert et al., 2010), the intestines were fixed and the positions of progeny analyzed in thick sections. Consistent with our organoid imaging, we observed that a subset of progeny (18/40 progeny pairs, n = 3 mice) were interspersed with unlabeled cells or differently labeled cells in the intact intestine (Figure 1E). Thus, progeny intersperse with neighboring cells in intestinal organoids and in the intestinal epithelium in vivo.

Cells intersperse during cytokinesis as part of a suite of cell shape changes restricted to the basolateral surface by cell-cell contact

We next sought to characterize the cell behaviors that give rise to interspersion during cell division in the intestinal epithelium. We observed that mixing occurred as cells underwent cytokinesis on the apical surface of the epithelium, during which neighboring cells intruded within the ingressing cytokinetic furrow (Figure 1B, Video 2). First, mitotic cells displaced to the apical surface of the epithelium, and the dramatic reduction in their basal footprint caused neighboring cells to reposition and occupy the position above (basal to) the mitotic cell (Figure 1B, Figure 1—figure supplement 1B). Cells progressed through a polarized (non-concentric) cytokinesis (Figure 2A, Video 2, Figure 2—videos 1, 2 and 3) (also see [Fleming et al., 2007]), in which the cleavage furrow initiated from the basal surface and then progressed to the apical surface. As cytokinesis continued, a minimal daughter-daughter contact remained on the apical surface (Figure 1—figure supplement 1E). We note that this minimal vertex contact is consistent with other reports of daughter cell geometry during vertebrate cytokinesis (Higashi et al., 2016), but contrasts with the long daughter-daughter interface generated during cytokinesis in Drosophila epithelia (Gibson et al., 2006; Herszterg et al., 2013; Pinheiro et al., 2017), as we will return to in the Discussion. The minimal contact between daughters generated by cytokinesis allowed a neighboring interphase cell to wedge between the daughters (Video 2). Finally, as the division completed, the daughter cells elongated on either side of the invading neighbor cell to occupy the full apical-basal axis in interphase (Figure 1, Video 2).

Figure 2 with 14 supplements see all
Polarized actin-dependent cell shape changes underlie division-coupled interspersion behaviors.

(A) Frames from time-lapse imaging of cytokinesis in an organoid expressing myosin regulatory light chain (MRLC)-mScarlet. (B) 3D reconstruction from live imaging of a cell dissociated from EB3-GFP organoids undergoing cytokinesis. EB3-GFP labeled organoids were used to facilitate identification of dissociated cells undergoing mitosis. Representative of 12/15 divisions. (C) Frames from SPIM of chromosome segregation in a live organoid. DNA: H2B-mScarlet. Arrowheads indicate mitotic chromosome masses. (D) Frames from confocal imaging of mitotic cells in live organoids treated with cytoskeletal inhibitors for 30 min before initiation of imaging. Membranes: R26mTmG; DNA: SiR-DNA. Arrowheads: mitotic chromosomes. (E) Quantification of the distance of mitotic chromosomes from the apical surface of the organoid epithelium following treatment with cytoskeletal inhibitors, normalized to the total apical-basal height of the epithelium, n ≥ 10. ns: not significant; ***p<0.001, Student’s t-test. (F) Anaphase of mitoses shown in (D). Dashed lines underline anaphase chromosome masses. (G) Frames from time-lapse imaging of Vil1CreERT2; R26mTmG organoids in which recombination has been induced at low levels to label a subset of cell membranes in the organoid. The protrusive front of one daughter cell is indicated by an arrowhead. Note that the division occurred along the imaging plane, such that the other daughter cell is ‘behind’ the imaged daughter cell. Asterisk: nearby interphase cell that did not participate in the division. (H) Frames from confocal imaging of live organoids testing the cytoskeletal requirements for the basal movement of nascent nuclei (top, arrowheads indicate chromosomes) and elongation of the basal cell edge (bottom, arrowhead indicates basal edge of reinserting cell). A schematic of this experiment is shown in Figure 2—figure supplement 1I. DNA: SiR-DNA; Membrane: R26mTmG; STLC: Eg5 inhibitor to induce mitotic arrest; SAC: spindle assembly checkpoint. (I) Quantification of DNA position before SAC inhibition (starting position), and at chromosome decondensation (end position), normalized to the total apical-basal distance of the epithelium. Arrowheads point towards the end position after mitotic exit. n ≥ 5, ns: not significant, ***: p<0.001, Student’s t-test of comparing end position and start position. Scale bars, 10 µm.

https://doi.org/10.7554/eLife.36739.009

In contrast to the dramatic shape changes on the basal surface of dividing cells, the apical surface remained unperturbed: the apical footprint of the mitotic cell was similar to its interphase neighbors (Figure 2—figure supplement 1A–C), and a cytokinetic furrow was absent from the apical surface as in many metazoan epithelia (Fleming et al., 2007Guillot and Lecuit, 2013Herszterg et al., 2013Founounou et al., 2013). Previous studies showed that cell-cell junctions on the apical surface of the intestine persist throughout mitosis (Jinguji and Ishikawa, 1992) and staining with junctional markers indicated that the same is true for intestinal organoids (Figure 2—figure supplement 1A). To test the possibility that persistent cell-cell contacts oppose mitotic shape changes on the apical surface, we dissociated organoids into single cells or pairs of cells and performed time-lapse imaging of mitotic exit. In contrast to the polarized cytokinesis that occurs in the tissue, cytokinesis occurred symmetrically in dissociated cells (Figure 2B, Figure 2—video 4), suggesting that tissue architecture plays a crucial role in this polarization. Together, these data indicate that mixing arises during cytokinesis as part of a suite of mitotic cell shape changes that are confined to the basolateral surface within the context of the tissue.

Rearrangements of the actin cytoskeleton during cell division displace dividing cells along the apical-basal axis

Our observations suggested that a critical initiating step during cell interspersion was the positioning of the dividing cell on the apical surface of the epithelium. We therefore sought to determine the mechanism that gives rise to this apical displacement. Apical displacement initiated concurrently with mitotic entry (Figure 2C, Figure 2—figure supplement 1D, Video 3 and Figure 2—video 5), indicating that it was distinct from interkinetic nuclear migration, a process in which the nucleus is moved apically during interphase (interkinesis) (Sauer, 1936) by actin or microtubule-based forces (reviewed in [Norden, 2017]). Apical displacement occurred as cells adopted the rounded geometry classically associated with mitosis (reviewed in [Théry and Bornens, 2008]) (Figure 2—figure supplement 1D, Video 1, Figure 2—video 5, Figure 2—video 6); at metaphase and anaphase, only fine membranous processes tethered the cell to the basal surface (Figure 2—figure supplement 1E–F), consistent with previous observations (Carroll et al., 2017; Fleming et al., 2007; Jinguji and Ishikawa, 1992; Trier, 1963). Mitotic rounding also contributes to late stages of interkinetic nuclear migration in some systems (Meyer et al., 2011; Spear and Erickson, 2012). Therefore, we tested the importance of actin-driven mitotic rounding for apical displacement. Treatment with the actin depolymerizing small molecule Latrunculin A disrupted rounding and apical displacement (Figure 2D,E, Figure 2—video 7); in contrast, cells treated with the microtubule depolymerizing drug nocodazole rounded onto the apical surface similarly to control cells (Figure 2D and E, Figure 2—video 8, Figure 2—video 9). As Latrunculin-treated cells entered anaphase, the chromosome masses were positioned orthogonally to the plane of the epithelium, in contrast to the planar divisions observed in control cells (Figure 2F, Figure 2—figure supplement 1G, Figure 2—videos 7, 9). This suggests that cell rounding is crucial for the normal planar orientation of the spindle in the intestine, as in some Drosophila epithelia (Chanet et al., 2017; Nakajima et al., 2013). Collectively, our data suggest that actin-based cell rounding displaces mitotic cells apically and is required for planar spindle orientation.

Video 3
Chromosome movements in intestinal organoids.

H2B-mScarlet labeled organoids were imaged by SPIM using 40X objectives at 2 min time points.

https://doi.org/10.7554/eLife.36739.024

We next assessed the mechanisms that restore the basal footprint and the basal position of the nuclei after division. After division, we observed that the basal edge of nascent daughters extended a protrusive front that resembled the leading edge of migrating cells (Figure 2G; Figure 2—video 10). Therefore, we tested the contributions of the actin cytoskeleton for basal reinsertion. As actin disruption blocks the initial displacement of mitotic cells to the apical surface (Figure 2D and E), determining the requirements for actin in basal reinsertion required that mitotic cells be positioned on the apical surface before disrupting actin. To achieve this, we first blocked cells on the apical surface by arresting them in mitosis with the mitotic kinesin (Eg5) inhibitor S-trityl-L-cysteine (STLC). Cells arrested in mitosis did not reinsert unless mitotic exit was induced by inhibition of the spindle assembly checkpoint (SAC; Mps1 inhibitor AZ3146) or cyclin-dependent kinase (CDK; RO-3306) (Figure 2—figure supplement 1H, Figure 2—video 11). Thus, mitotic exit and reversal of CDK phosphorylation are sufficient for basal reinsertion, even in the absence of chromosome segregation.

Using this mitotic arrest and exit protocol, we tested the requirements for the actin and microtubule cytoskeletons for basal reinsertion (Figure 2—figure supplement 1I). When we disrupted the actin cytoskeleton and induced mitotic exit, the nucleus reformed its interphase morphology on the apical surface and the cell boundary did not protrude toward the basal surface (Figure 2H,I, Figure 2—video 12). In contrast, depolymerizing microtubules with nocodazole and inducing mitotic exit did not interfere with the ability of nuclei or the cell boundary to reach the basal surface (Figure 2H,I, Figure 2—video 13). Although actin also plays a critical role in cytokinesis, nuclei reinserted normally following inhibition of cytokinesis using the Polo-like kinase one inhibitor, BI2536 (Lénárt et al., 2007; Steegmaier et al., 2007) (Figure 2—figure supplement 1J), indicating that cytokinesis is dispensable for basal movement. Collectively, these data indicate that actin-driven cell elongation after mitotic exit re-establishes the interphase architecture of daughter cells.

The cellular aspect ratio is a key parameter for allowing interspersion during division

Our data indicate that the displacement of cells along the elongated apical-basal axis over the course of cell division plays a role in cell interspersion. To test the importance of an elongated apical-basal axis for cell interspersion, we imaged cell behavior in spherical organoids derived from fetal intestine (Fordham et al., 2013; Mustata et al., 2013), in which cells are very short in the apical-basal dimension, and are instead elongated along the sphere circumference (Figure 3A, Figure 3—video 1). Fetal spheroids did not exhibit apical-basal mitotic movements and the daughters did not intersperse with other cells during division (Figure 3A,B, Figure 3—video 1, Figure 3—video 2) (50/50 divisions).

Figure 3 with 4 supplements see all
Cellular aspect ratio is a key parameter for division-coupled interspersion in the intestine and the embryo.

(A) 3D reconstruction of a spherical organoid derived from fetal (E13.5) mouse intestine. DNA: Syto 21 dye, arrowhead: mitotic cell. (B) Frames from 3D reconstructed time-lapse SPIM of chromosome segregation in spheroids cultured in exogenous Wnt3a (left, DNA: H2B-GFP) or derived from fetal intestine (right, DNA: H2B-mScarlet). Arrowheads: dividing cell. Views from the basal surface are shown. (C) Immunofluorescence images of secretory-lineage labeled organoids (Atoh1CreER; R26RFP) grown with buds (left) or as spheres (right) in exogenous Wnt3a. Actin: Alexa 488-phalloidin, DNA: Hoechst 33342. Image scaled with ɣ adjustment. (D) Quantification of the frequency of division-coupled interspersion in three replicates (top, n = 20 divisions per replicate) and cellular aspect ratio (bottom, n = 20). (E) Cartoon depicting the embryonic portion of an E7.5 (late streak) mouse egg cylinder, distal end up. (F) Frames from 3D reconstructed time-lapse SPIM of stochastically labeled cells (RFP + cells of CAGGSCreER; R26Brainbow2.1) in the epiblast/primitive ectoderm. Arrowhead: cell displacing to the apical surface as it enters mitosis. (G) 3D reconstruction of H2B-GFP embryos. Arrowheads: mitotic chromosomes. (H) Frames from 3D reconstructed time-lapse SPIM of stochastically labeled cells (RFP + cells of CAGGSCreER; R26Brainbow2.1) in the epiblast/primitive ectoderm. Arrowheads: dividing cell and nascent daughters, which become separated by an unlabeled cell. Scale bars, 10 µm.

https://doi.org/10.7554/eLife.36739.025

We also induced a subset of adult intestinal organoids to adopt a spherical geometry and short apical-basal axis by addition of exogenous Wnt to the medium (Sato et al., 2011) (Figure 3—figure supplement 1A). These adult spheroids also failed to exhibit apical-basal mitotic movements and the daughters did not intermix with other cells (50/50 divisions) (Figure 3B, Figure 3—figure supplement 1A). Consistent with the lack of interspersion, these spheroids contained patches of cellular progeny (Figure 3C), in contrast to the interspersed pattern of cell lineages observed in normal adult organoids (Figure 1A). As an internal control, a subset of organoids cultured in high Wnt conditions retained their budded morphology and elongated apical-basal cell shape; these organoids continued to exhibit apical displacement and the interspersed pattern of cell lineages (Figure 3C). This experiment, as well as our observations of adjacent progeny in the fetal spheroids, which exhibit very low expression of the Wnt reporter gene Axin2 (Mustata et al., 2013), indicate that the effect of cell shape on interspersion is separable from hyperactive Wnt signaling, in contrast with previous work (Carroll et al., 2017). Together, these data indicate that an elongated apical-basal axis is critical for apical mitosis and cell interspersion during division.

Apical displacement during division underlies cell interspersion in the elongated epithelium of the embryonic primitive ectoderm

Based on our data suggesting a crucial role for the cellular aspect ratio in interspersion in the organoids (Figure 3D), we next examined whether the mechanisms that we defined in the intestine may be relevant to other tissues with similar physical parameters. Pioneering work by Gardner and Cockroft (1998) revealed that cells injected into mouse blastocysts to generate chimeras become dispersed throughout the epiblast and primitive ectoderm of the post-implantation embryo. The authors proposed that this pattern might arise as a consequence of cell division, which they and others have observed occurs on the apical surface of the tissue (Gardner and Cockroft, 1998; Ichikawa et al., 2013). Therefore, we tested this prediction by performing time-lapse SPIM imaging of E7.5 (late streak-early bud) mouse embryos (Figure 3E), in which the epiblast/primitive ectoderm was mosaically labeled (CAGGSCreER; R26Brainbow2.1). We imaged cell divisions in these embryos for at least 3 hr and observed that divisions proceeded in a similar manner to the intestinal epithelium, with mitotic cells displacing to the apical surface as they rounded (Figure 3F–G). Daughter cells then separated from one another and interspersed with unlabeled cells during cytokinesis (Figure 3H, Video 4) (8/10 divisions, n = 3 embryos from three pregnancies). Thus, daughter cells positioned on the apical surface intersperse with other cells during cytokinesis in the elongated epiblast/primitive ectoderm of the embryo, as in the adult small intestine. In contrast, the cells of the visceral endoderm (the low aspect ratio cells that surround the epiblast) did not exhibit apical displacement and daughters remain adjacent (Figure 3—figure supplement 1B, Figure 3—video 3, 12/12 divisions, n = 3 embryos from three pregnancies), consistent with classical experiments reporting outgrowth of contiguous clones in this tissue (Gardner, 1984; 1985; Lawson et al., 1991). Thus, cell division generates distinct progeny patterns in the two layers of the early post-implantation mouse embryo, consistent with a central role for cellular aspect ratio in determining the spatial patterning of cell progeny.

Video 4
Cell separation during division in the embryonic epiblast/primitive ectoderm.

An E7.5 mouse embryo was imaged by SPIM with 40X objectives at 4 min time points. Cells expressing cytoplasmic RFP from a CAGGSCreER; R26Brainbow2.1/+ embryo are shown.

https://doi.org/10.7554/eLife.36739.030

Discussion

The functions of epithelial organs rely on the concerted action of multiple cell types. As these cell types are replenished as the organ renews, they must be positioned appropriately within the tissue. In some mammalian epithelia, such as the small intestine, daughter cells derived from a common progenitor disperse throughout the tissue and intermingle with cells of other lineages, a process that plays an important role in determining local signaling environments. Previous studies have reported that intermingling of cells can occur during cell division (Carroll et al., 2017; Firmino et al., 2016; Gardner and Cockroft, 1998; Higashi et al., 2016; Lau et al., 2015; Packard et al., 2013) but the mechanism by which this occurs has not been clear. Here, we show that intermixing arises when a neighboring cell inserts between apically displaced daughter cells during cytokinesis.

The process of intermixing requires that the neighboring cell and dividing cell are positioned in such a way that the neighbor can occupy the wedge between the daughters generated by the ingressing furrow. Our data support a model in which the neighboring cell can become opportunely positioned for invasion into the cytokinetic furrow as a consequence of the cell shape changes associated with vertebrate mitosis in tissues comprised of cells with a high aspect ratio (Figure 4). In cells with a high aspect ratio, the actin-driven cell shape changes required for mitosis (rounding and subsequent elongation) displace the dividing cell along the apical-basal axis (Figure 4). As a result, an elongated interphase neighboring cell can surround the dividing cell both basally and laterally, allowing it to follow the path of the ingressing furrow between the daughters. Consistent with a key role for cell aspect ratio in interspersion behavior, reducing the aspect ratio in organoids generates patches. Live imaging of cell division in the two epithelial layers of the peri-gastrulation mouse embryo further supports a model in which cell aspect ratio is a critical parameter for determining whether cellular progeny intersperse, raising the intriguing possibility that the patterning principles that we define in the intestine may be a common feature of many mammalian epithelia.

Figure 4 with 1 supplement see all
Model for cell progeny patterning in mammalian epithelia.

Top: Cartoon of the influence of cell height on the relative positioning of cells derived from a given progenitor (magenta cells). Bottom: Model for interspersion of cell progeny in elongated epithelia. The basolateral surface of a dividing cell undergoes dramatic actin-dependent changes in cell shape that displace the chromosomes and cell body along the apical-basal axis. Neighboring cells can position within the ingressing cytokinetic furrow, displacing daughter cells from one another as they reinsert onto the basal surface.

https://doi.org/10.7554/eLife.36739.031

Several lines of evidence support a model in which interspersion arises as a mechanical consequence of executing planar cell division in elongated cells, rather than being determined by developmental signaling or differential adhesion between cells. First, daughter separation is observed throughout the intestinal crypt for all progenitor cell identities: stem cells, Notch-expressing absorptive progenitors and Atoh1-expressing secretory progenitors (Figure 1Figure 1—figure supplement 1A). However, importantly, daughter separation is a frequent but not universal event, occurring in approximately half of the divisions observed, including when observing cells of a specific lineage (Figure 1Figure 3D). Additionally, altering cell fates, for example by inhibiting Notch signaling to cause an expansion of secretory cells, does not alter the frequency of this process (Figure 1—figure supplement 1C,D). In contrast, altering epithelial geometry in culture disrupts interspersion (Figure 3).

Since our data indicate that interspersion can arise from the execution of planar cell division coupled with the physical parameters of the tissue, it raises the possibility that the mechanisms of interspersion that we define for the intestinal epithelium may be generalizable to other vertebrate tissues with similar physical parameters. Consistent with this notion, we observed similar interspersion in the high aspect ratio epithelium of the early mouse embryo, while the surrounding low aspect ratio epithelium did not exhibit division-coupled interspersion (Figure 3). Several tissues across vertebrates with a high aspect ratio have also been reported to exhibit division-coupled interspersion (Carroll et al., 2017; Firmino et al., 2016; Gardner and Cockroft, 1998; Higashi et al., 2016; Packard et al., 2013). In contrast, in numerous tissues in which cells have a low aspect ratio, progeny remain adjacent and form contiguous patches, including the interfollicular epidermis (Ouspenskaia et al., 2016; Rompolas et al., 2016), MDCK cells (Reinsch and Karsenti, 1994), and alveolar epithelial cells (Desai et al., 2014). Our model raises the possibility that isolated reports of division-coupled interspersion in diverse vertebrates including frog, chick and mouse may be unified by a common physical mechanism arising from the aspect ratio of the tissue and the mechanics of cell division.

While our data indicate that cellular aspect ratio is an important parameter for interspersion, the mechanics and geometry of cytokinesis also appear to play a central role. In vertebrates, the mechanism of furrow ingression minimizes the contact between the daughters and progresses until a single apex physically connects the two cells (Higashi et al., 2016) (Figure 1—figure supplement 1E, Figure 4—figure supplement 1). An important component of our model is that the development of the furrow creates a position, both basally and laterally, for neighboring cells to invade and occupy. However, in contrast, during cytokinesis in Drosophila, the two daughters form a long adhesive contact between them (Gibson et al., 2006) (Figure 4—figure supplement 1), dependent on myosin II accumulation in the neighboring cells (Herszterg et al., 2013; Pinheiro et al., 2017). In this regard, it is interesting to note that Drosophila epithelia exhibit a high aspect ratio, apical mitosis and non-concentric cytokinesis, yet do not exhibit cell interspersion and form contiguous patches of progeny (Bryant, 1970; Bryant and Schneiderman, 1969; Founounou et al., 2013; Gibson et al., 2006; Guillot and Lecuit, 2013; Herszterg et al., 2013; Meyer et al., 2011; Morais-de-Sá and Sunkel, 2013). We speculate that the extended cell-cell contact formed between daughter cells in Drosophila would oppose the invasion of a neighboring cell. In the future, it will be interesting to attempt to modify the extent of interactions between daughter cells either in Drosophila or vertebrate epithelia and determine the effects on progeny patterning.

Broadly, since our data suggest that cell interspersion requires a set of criteria that are satisfied by many vertebrate epithelia, it is unlikely to be unique to those tissues in which it has been reported. Although our work has focused on the columnar epithelium of the small intestine, in which mitotic cell shape changes are sufficient to displace dividing cells relative to their neighbors, the numerous elongated pseudostratified epithelia that undergo apical mitosis due to interkinetic nuclear migration (reviewed in [Norden, 2017]) are particularly attractive candidates for division-coupled interspersion. Together, our model suggests that interspersion during cell division may be widespread across elongated vertebrate epithelia.

Materials and methods

Key resources table
Reagent type
(species) or
resource
DesignationSource or referenceIdentifiersAdditional information
Strain, strain
background
(Mus musculus)
R26mTmGJackson Labs,
PMID: 17868096
MGI: 3716464
Strain, strain
background
(Mus musculus)
Vil1Cre-ERT2Averil Ma lab,
PMID: 15282745
MGI: 3053826
Strain, strain
background
(Mus musculus)
Atoh1CreERTJackson labs,
PMID: 16958097
MGI: 3686985
Strain, strain
background
(Mus musculus)
R26RFPJackson Labs,
PMID: 20023653
MGI: 3809524
Strain, strain
background
(Mus musculus)
Lgr5DTR-GFPde Sauvage Lab
(Genentech),
PMID: 21927002
MGI: 5294798
Strain, strain
background
(Mus musculus)
C57BL/6JJackson Labs
Strain, strain
background
(Mus musculus)
R26Brainbow2.1Jackson Labs,
PMID: 20887898
MGI: 164644
Strain, strain
background
(Mus musculus)
H2B-GFPJackson Labs,
PMID:15619330
MGI: 109836
Strain, strain
background
(Mus musculus)
Notch1CreERT2(SAT)PMID: 21991352MGI: 5304912
AntibodyRabbit anti-ZO-1Thermo FisherRRID:AB_2533456
Chemical
compound, drug
Alexa 488-PhalloidinThermo FisherA12379
Chemical
compound, drug
Hoechst 33342Molecular ProbesH3570
Chemical
compound, drug
See Table 1 for
pharmacological inhibitors
Software,
algorithm
MicroManagerOpen Imaging,
PMID: 20890901

Mouse strains and lines

Adult mice of the following lines were used to generate organoids.

R26mTmG/mTmG (Muzumdar et al., 2007) (female)

Vil1Cre-ERT2/+ (el Marjou et al., 2004); R26mTmG/+ (male)

Atoh1CreERT/+ (Chow et al., 2006); R26RFP/+ (Madisen et al., 2010); Lgr5DTR-GFP/+ (Tian et al., 2011) (female)

Notch1CreERT2 (SAT)/+ (Fre et al., 2011); R26RFP/ RFP (Madisen et al., 2010) (female)

Fetal organoids were generated from E13.5 C57BL/6J embryos.

For imaging of cell interspersion in the intact intestine, adult Vil1Cre-ERT2/+ (el Marjou et al., 2004); R26Brainbow2.1/+ (Snippert et al., 2010) mice were used. Recombination was induced by oral gavage with one dose of 2.5 mg tamoxifen in corn oil 3 days before analysis.

Brainbow embryos were generated by crossing CAGGSCreER/+ males (Hayashi and McMahon, 2002) to R26Brainbow2.1/Brainbow2.1 (Snippert et al., 2010) females. Plugged females were injected intraperitoneally with 2.5 mg tamoxifen in corn oil at E5.5. H2B-GFP embryos were generated by crossing H2B-GFP males (Hadjantonakis and Papaioannou, 2004) to C57BL/6J females. Embryos were dissected at E7.5 and staged according to (Delling et al., 2016; Downs and Davies, 1993).

The strains of these mice were the same as previously described in their respective references at the time of acquisition but were subsequently maintained on mixed backgrounds after breeding between different lines. All experiments involving mice were approved by the Institutional Animal Care and Use Committee of the University of California, San Francisco (protocol #AN151723).

Organoid preparation, dissociation and immunofluorescence

Small intestinal crypts were isolated from adult mice or E13.5 embryos and cultured in medium supplemented with human recombinant EGF, human recombinant Noggin and R-Spondin conditioned medium (ENR medium) as described (Sato et al., 2009). Catalog numbers for culture medium components are described in (Mahe et al., 2013). R-spondin and Wnt3a conditioned medium were used where indicated. Lentiviral transduction of adult organoids was performed as described (Koo et al., 2011). Fetal organoids were transduced according to the same protocol, but without the addition of exogenous Wnt3a to the medium at any step. For propagation, organoids were grown in 24-well plastic plates. For spinning disc imaging and immunofluorescence, organoids were grown in 96-well glass bottom dishes (Matriplate, Brooks). For SPIM, organoids were grown on glass coverslips which were then transferred to the SPIM imaging chamber (see below). For immunofluorescence, organoids were fixed in 4% PFA in PBS for 1 hr before blocking in 3% BSA, TBS, 0.1% Triton X-100. Primary antibody was incubated overnight at four degrees and secondary antibody was incubated for >2 hr at RT. Reagents used for immunofluorescence were as follows: rabbit anti-ZO-1 antibody (Thermo Fisher), Alexa488-Phalloidin (Thermo Fisher # A12379), Hoechst 33342 (Molecular Probes H3570).

For organoid dissociation, organoids in one well of a 24 well plate were washed once in PBS before Matrigel was manually disrupted by pipetting in TrypLE Select (Life Technologies) in the well. The plate was then incubated at 37°C for 7–8 min before additional disruption with a P200 pipette. The cell suspension was centrifuged in medium +5% fetal bovine serum at 1000 x g for 5 min. The pellet was resuspended in Matrigel, allowed to polymerize for 10 min and covered with ENR medium and immediately transferred to the microscope for imaging for 45 min – 1 hr.

Tissue preparation for clone tracing

Animals were anesthetized by intraperitoneal (i.p.) injection of 250 mg/kg of body weight avertin (2,2,2-tribromoethanol) and transcardially perfused with 4% paraformaldehyde (PFA) in 0.1 M phosphate-buffered saline (PBS). Dissected tissues were post-fixed in 4% PFA for 3 hr at 4°C and cryoprotected in 30% sucrose in 1 × PBS overnight at 4°C. For whole mount tissue, the external smooth muscle and fat of the most proximal 3 cm of the small intestine was removed and epithelial tissue was coverslipped with ProLong Gold Antifade (P36930, Thermo Fisher Scientific). For tissue sections, tissue was embedded in OCT compound (4583, Sakura), frozen and stored at −80°C. Small intestine swiss rolls were cryosectioned at 50 µm and coverslipped with ProLong Gold Antifade. Whole mount tissue and sections were counterstained with DAPI (1:10000; D9542, Sigma) for 45 min or 15 min, respectively.

Microscopy

For spinning disc confocal imaging, images were acquired on a Yokogawa CSU-X1 spinning disk confocal attached to an inverted Nikon TI microscope, an Andor iXon Ultra 897 EM-CCD camera, using Micro-Manager software (Edelstein et al., 2010). Imaging of 12 × 1 µm z-stacks was performed either at 4 min time intervals with a 40 × 1.30 NA Plan Fluor oil objective or a 20 × 0.75 NA objective, or at 20 s time intervals with a 60XA 1.20 NA Plan Apo water immersion objective. Maximum intensity projections of 1–5 Z-stacks are shown unless otherwise noted. Point-scanning confocal imaging of intact intestines was performed using a Leica TCS SP8 X confocal microscope, with HyD and LAS X software. 0.76 μm optical sections were acquired sequentially with a 63 × 1.40 HC PL APO CS2 oil objective.

4-dimensional imaging was performed on an ASI diSPIM microscope equipped with 40 × 0.80W NA NIR-Apo water dipping objectives, Hamamatsu Flash 4.0 cameras, and 488 nm and 561 nm solid state lasers from Vortran, using a nightly build of the Micro-Manager software. The structure of the environmental control chamber is described in detail at https://valelab4.ucsf.edu/~nstuurman/protocols/diSPIMIncubator/. Temperature was maintained using 3 × 50 ohm resistors attached to the stainless steel incubation chamber holding the coverslip and medium, a 10 kOhm thermistor inserted in the medium and a temperature controller (TE Technology, Inc. TC-48–20). O2 and CO2 tensions in the medium were kept constant by flowing humidified gas underneath the sample chamber. To allow gas exchange, the sample was placed on a sandwich of 2 × 24 ×50 mm coverslip glasses in which 2 ~ 12×12 mm windows had been laser-cut and between which a piece of ~37.5 µm thick Teflon AF-2400 (a gift from BioGeneral, Inc.) was placed. Evaporation was minimized by layering mineral oil (Howard) over the sample. Organoids were imaged in ENR medium; embryos were imaged in DMEM +25% rat serum (Rockland, Inc.). 3D reconstructions were generated using a Micro-Manager plugin (https://github.com/nicost/MMClearVolumePlugin) that uses the ClearVolume library (Royer et al., 2015). 3D reconstructions are scaled with gamma adjustment. All imaging experiments were performed at 37°C, 5% CO2, 20% O2.

Small molecules

Small molecule concentrations are described in Table 1. All stock solutions were prepared in DMSO. All pharmacological experiments were performed in the presence of 10 µM Verapamil to inhibit drug efflux.

Quantification and statistical analysis

Details of statistical tests are provided in the figure legends. A statistical method of sample size calculation was not used during study design. Data were pooled from at least three biological replicates. When the observations presented were observed in less than 100% of cases, their frequency is noted in the figure, figure legend and/or text.

Table 1
Small molecules used in this study.
https://doi.org/10.7554/eLife.36739.033
MoleculeFunctionSourceCat #Final concentration
 NocodazoleMicrotubule inhibitorCalbiochem4879295 µM
Latrunculin AF-actin inhibitorCalbiochem4280264 µM
SiR DNADNA dyeCytoskeleton IncCY-SC0071 µM
VerapamilEfflux pump inhibitorCytoskeleton IncCY-SC00710 µM
MG132Proteasome inhibitorSigmaML44910 µM
STLCEg5 inhibitorSigma16473910 µM
RO-3306CDK inhibitorCalbiochem21769910 µM
AZ3146Mps1 inhibitorTocris39942 µM
BI2536Plk1 inhibitorSelleck BiochemS110910 µM
TamoxifenCre-ER inducer
(applied for 6–16 hr in culture)
SigmaT56481 µM
(Atoh1CreER and Notch1CreER)
0.1 µM
(Vil1CreER)
S -- BlebbistatinMyosin II inhibitorAbcamab120491200 µM
S -- BlebbistatinMyosin II inhibitorCayman13013200 µM
Y27632ROCK inhibitorSelleckS104910 µM
DAPTGamma-secretase (Notch) inhibitorAbcamab12063350 µM

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
    Staging of gastrulating mouse embryos by morphological landmarks in the dissecting microscope
    1. KM Downs
    2. T Davies
    (1993)
    Development 118:1255–1266.
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
    Complete dissipation of coherent clonal growth occurs before gastrulation in mouse epiblast
    1. RL Gardner
    2. DL Cockroft
    (1998)
    Development 125:2397–2402.
  19. 19
    An in situ cell marker for clonal analysis of development of the extraembryonic endoderm in the mouse
    1. RL Gardner
    (1984)
    Journal of Embryology and Experimental Morphology 80:251–288.
  20. 20
    Clonal analysis of early mammalian development
    1. RL Gardner
    (1985)
    Philosophical Transactions of the Royal Society B: Biological Sciences 312:163–178.
    https://doi.org/10.1098/rstb.1985.0186
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
    Clonal analysis of epiblast fate during germ layer formation in the mouse embryo
    1. KA Lawson
    2. JJ Meneses
    3. RA Pedersen
    (1991)
    Development 113:891–911.
  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

Decision letter

  1. Jody Rosenblatt
    Reviewing Editor; University of Utah, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your work entitled "Cellular aspect ratio and division mechanics govern the patterning of cell progeny in mammalian epithelia" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, including Jody Rosenblatt as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by a Senior Editor.

Our decision has been reached after consultation between the reviewers. All the reviewers found the study interesting and compelling in its nature, given that this is an important, fundamental finding that is not well understood. However, there was general concern that the conclusions were oversold regarding the idea that tall epithelia intrinsically drive interspersion of cells. The results presented did not provide a clear mechanism for how the interdispersion takes place following cytokinesis, given that interspersion does not take place in other tall epithelia. The height of epithelia seemed correlative, using a small n for different types of epithelia or whether other epithelial cell types in the same organoid also disperse. Moreover, other in vivo components may contribute to dispersion that are lacking in the organoids; for instance the crypt curvature, which is not as well maintained in organoids, may contribute to the essential architecture required for interspersion. If so, the authors may gain more insight by filming divisions in ex vivo gut slices as well.

Because we felt that better supporting the claims stated in the title would take longer than the typical revision period of two months, the reviewers decided that it would be best turn down this manuscript. While I have summarized the main critiques of the paper, below, we include the full reviews.

Reviewer #1:

This beautifully written paper by McKinley and others reveals a new mechanism for how two daughter cells become interspersed through an epithelium. They show that interspersion occurs by cytokinesis at the apex of the epithelium and then re-attachment of two daughter cells basally to the matrix after they straddle a cell between them. They show that this mechanism is dependent on having a taller epithelium and that it also requires actin post-cytokinesis for cells to straddle and stretch down basally. I recommend publication of this manuscript that provides mechanistic insight to a long-observed puzzle in epithelial morphology, which is how do certain types of cells become interspersed throughout a particular epithelium. I have only a small number of questions that would be good to address:

1) The correlation between columnar epithelial height and the propensity to cause interspersion of daughter cells looks compelling but lack graphical analysis. It seems like a perfect place to include this would be in Figure 3C. Here, they could show a graphical correlation on cell height with wnt-treated organoids and the number of interspersed divisions or cells. It may be in the text but is not easy to see graphically and such a graph could make the point very clearly.

2) One thing that was not clear is whether all the cells within intestinal epithelia are interspersing or is it just the secretory lineage. This is important once we later find that the interspersion is linked to cell height. If so, we would expect all cells within the intestinal epithelium to intersperse. This would also predict that mitotic forces expected to drive cell migration up the villus will be dispersed, I expect.

3) Did the flatter visceral endoderm not intersperse in your videos? This would also lend more credence to the height as a controller.

Reviewer #2:

This manuscript examines how cell-types become interspersed during the development of intestinal organoids and presents evidence that this occurs as cells undergo cytokinesis at the apical surface and are then displaced from their siblings as they reintegrate towards the basal surface. This is an interesting question that takes advantage of the tractability of organoids for cell biological and live imaging approaches. While the manuscript is a useful contribution, I am not convinced that the data fully justify their conclusions. Most of the behaviours they observe, such as the apical movement and rounding of mitotic cells and planar spindle orientation have been well documented in a variety of other columnar or pseudostratified epithelia, so the main novelty lies in the mechanism of cell separation at cytokinesis. While their description is presumably accurate, it falls short of a mechanistic explanation and fails to consider the roles of the non-dividing basal neighbours or how cell junctions are remodelled during this process. As pointed out in the Discussion, divisions in the pseudostratified epithelia of the Drosophila imaginal discs share all of the features of the divisions in adult-derived intestinal organoids, including the polarised ingression of the cleavage furrow from basal to apical, yet the daughter cells are only very rarely separated during this process. Thus, any valid mechanistic model of cell dispersal in the intestine should be able to explain the Drosophila imaginal disc is different and the data as presented do not show any significant differences. The difference may lie in the behaviours of the neighbouring cells, as recent work has shown that the neighbours in the disc activate myosin contractility to pull on the apical junctions of the dividing cell during cytokinesis, which ensures that a new cell-cell junction forms between the newly-born daughters (Pinhiero et al., 2017). I therefore think that these results, useful though they are, have not really solved the question of cell dispersal and the conclusions are over-interpreted. The results are a valuable contribution to the field, and this might be publishable if the conclusions were less over-blown. For example, they could concentrate on their observations that the apical-basal height of the epithelium is a key parameter for allowing dispersal.

The authors claim that cell-cell junctions on the apical surface promote the polarized, asymmetric mitotic cell-shape changes (Subsection “Cells intersperse during cytokinesis as part of a suite of cell shape changes restricted to the basolateral surface by cell-cell contact”/Figure 2B/Figure 2—figure supplement 1A-E). For this, they measure the positioning of cortical bleb formation in mitotic cells within the tissue context and after dissociation into single cells. They show that blebbing occurs only basally in the tissue (Figure 2—figure supplement 1A) and argue that this is inhibited on the apical side by cell-cell contacts ("…cell-cell junctions on the apical surface…oppose mitotic shape changes on the apical surface"), as blebbing occurs symmetrically in dissociated cells. However, in my opinion, the proposed mechanistic link between cell-cell junction and asymmetric cell shape changes are insufficiently addressed by means of these experiments. For instance, if cell-cell junctions on the apical side have an inhibitory effect on bleb formation, why is blebbing constrained to the basal surface (Figure 2—figure supplement 1A) and does not occur on the lateral sides? Is it possible that the "fine membranous processes that tether the cell to the basal surface" (subsection “Rearrangements of the actin cytoskeleton during cell division displace dividing cells along the apical-basal axis”, first paragraph and Figure 2—figure supplement 1G-H) have never detached from the basement membrane when the cell body moved apically.

The dotted round cell outlines in Figures 2D, 2H, Figure 2—figure supplement 1L are confusing. It would be better to draw the outline of the cells based on the actual cell shape shown in the membrane signal.

Figure 2B and Figure 2—figure supplement 1E: It is not clear why the authors suddenly switch to a cell line with MT +end marker when they analyze membrane blebbing in dissociated cells? Why not stick to the cell line with a membrane marker as in previous experiments?

Reviewer #3:

The overarching question addressed in this study is that of how cells become interspersed within epithelia. This is a really interesting and universal open question with important implications. I was therefore really excited to get to read this paper, and unfortunately I need to admit I got a bit disappointed. The concluding statements made are bold, but the data presented, and more importantly system used for bulk of the analysis presented though quite clever, seems a little contrived. That said, this is not my field and I could easily be swayed.

The focus of this study is the mouse small intestine. Live (light-sheet and confocal etc.) imaging of small intestinal organoids in which subsets of cells are labelled in various genetic ways (inducible lineage-specific mosaic Cre lines, Confetti to mark clones etc.) is used to characterize the behavior of cells and begin to define mechanisms of cell interspersion within organoids as a proxy for the intestinal epithelium.

The authors use organoids derived from adult (tall) intestine and show that actin-, but not tubulin-, based behaviors drive cell interspersion, both cell rounding displacing cells apically at division and for subsequent re-establishment of a basal foothold. They then contrast these findings with fetal (short) intestinal-derived organoids which do not exhibit these behaviors (do they show this leads to a failure in interspersion and coherent clonal expansion within the epithelium?).

The authors nicely describe how cell interspersion arises when interphase neighboring cells wedge between apically dividing daughters during cytokinesis, and requires an elongated (tall) epithelium, and that when the cellular aspect ratio is perturbed interspersion is disrupted resulting in local clonal expansion (and the production of clonal patches).

Even though separation of daughter cells has been described previously in various tissues in vivo, it is not clear whether the mechanisms defined here in organoids are at play within the crypts/villi, and no in vivo correlations with the in vivo situation are made for any of the behaviors described for organoids. I feel this is a major weakness of the study.

https://doi.org/10.7554/eLife.36739.036

Author response

We are writing concerning our submission to eLife and asking that it be considered for re-review.

We believe that this is an important paper that adds much needed cell biology and mechanism to a fundamental problem (cell dispersion). The two main findings are that 1) daughter cell separation occurs at the point of cytokinesis (which differs from other thoughts in the field), and 2) that a difference in the cell aspect ratio between interphase and mitotic cells impacts whether daughters disperse or remain together to form clonal patches. More importantly, in the intervening time since the paper was submitted, we have performed experiments that we believe squarely address the main criticisms that emerged in the decision letter, as well as extensively revising the text in response to the feedback received.

Point 1:

All the reviewers found the study interesting and compelling in its nature, given that this is an important, fundamental finding that is not well understood. However, there was general concern that the conclusions were oversold regarding the idea that tall epithelia intrinsically drive interspersion of cells. The results presented did not provide a clear mechanism for how the interdispersion takes place following cytokinesis, given that interspersion does not take place in other tall epithelia.

As discussed below in point 2c, we have extensively revised the manuscript to temper our statements regarding the pertinence of our model to other mammalian systems that we have not tested.

The comment also centers around the fact that tall Drosophila epithelia have not been reported to exhibit interspersion. This is an important consideration that we did not address sufficiently in our original manuscript. Addressing this point has allowed us to add depth and precision to our model for interspersion, and we are grateful for the encouragement to make these changes. Our data indicate that interspersion arises from a neighboring cell invading between the daughters during cytokinesis. This invasion depends on the process of cytokinesis generating a position between the daughter cells for the neighbor to occupy. Thus, our model suggests that the creation of a minimal contact between the daughters is central to the interspersion behavior; extensive contact/adhesions between the daughters during cytokinesis would oppose the invasion of the neighbor. Importantly, extensive literature has demonstrated that Drosophila cytokinesis generates a long adhesive contact between the daughters (Gibson, Patel, Nagpal, and Perrimon, 2006; Herszterg, Leibfried, Bosveld, Martin, and Bellaiche, 2013; Pinheiro et al., 2017). By our observations and model, one would expect this geometry to oppose neighbor cell invasion. In contrast, a minimal vertex connection is created between daughters during cytokinesis in our system and broadly in vertebrates (analyzed carefully in (Higashi, Arnold, Stephenson, Dinshaw, and Miller, 2016), please also see our new figures – Figure 4—figure supplement 1 and Figure 1—figure supplement 1E). Thus, the neighboring cell can wedge between the daughters.

We apologize for our superficial discussion of this important consideration in the previous draft, and we appreciate the opportunity to include these important refinements of our model, now included in several points in the text, and addressed carefully in the second to last paragraph in the Discussion. Naturally, the differences between cell interspersion in Drosophila and vertebrates begs for further experiments to test our hypothesis. This would be best accomplished by somehow modifying the extent of contact between daughters during cytokinesis. We describe this possibility in the paper as a possible future direction in the Discussion. However, this is clearly beyond the scope of our current study, both because of the complexity of the experiment and time to pioneer a completely new study of comparing and manipulating vertebrate and invertebrate systems.

Point 2:

The height of epithelia seemed correlative, using a small n for different types of epithelia or whether other epithelial cell types in the same organoid also disperse.

We have now performed the additional experiments requested by the reviewers to further support our model for the role of cell height, as well as rewriting the text to focus on the specific epithelia we have analyzed. Our paper now analyzes both epithelia of the embryo, all cell lineages in the organoid, as well as several alternative organoid systems (adult versus fetal, low and high Wnt conditions). The new additions are as follows:

a) Reviewers 1 and 3 suggested that we examine the visceral endoderm of the embryo. We have examined cell divisions in the visceral endoderm of the embryo using embryos in which the DNA is labeled with H2B-GFP. In contrast to the elongated epiblast/primitive ectoderm, the endodermal cells do not intersperse during division in this low aspect ratio tissue. The new figure 3—figure supplement 1B, and is accompanied by a new video (Figure 3—video 3).

b) We have now demonstrated that all cell lineages in the intestinal crypt exhibit interspersion, as expected based on our model indicating that aspect ratio, rather than e.g. cell identity, is a key parameter for this behavior. Previously, we only analyzed interspersion in the stem cell lineage (Lgr5-expressing) and the secretory (Atoh1-expressing) lineage. We have now analyzed Notch1-expressing cells, which represents all non-secretory cell populations, and found that these cells also intersperse at a similar frequency to secretory cells. Together, these reporters cover all cell lineages in the crypt epithelium. (New Figure 1—figure supplement 1A).

c) We have been careful to rewrite the text to state conclusions based on the subset of mammalian epithelia that we have directly tested. We have tempered our conclusions with regard to breadth in the title, abstract and throughout the text to make it clear that the principles that we have uncovered may pertain to several (not necessarily all) mammalian epithelia.

Point 3:

Moreover, other in vivo components may contribute to dispersion that are lacking in the organoids; for instance the crypt curvature, which is not as well maintained in organoids, may contribute to the essential architecture required for interspersion. If so, the authors may gain more insight by filming divisions in ex vivo gut slices as well.

We now demonstrate in vivo relevance of interspersion in the intestine. Following reviewer 3’s suggestion, we analyzed fixed intestines labeled for three days with the stochastic multicolor reporter (often referred to as confetti/Brainbow) to determine if interspersion of progeny is occurring in vivo. We found that cells derived from a given progenitor also intersperse with other cells in vivo (new Figure 1E). Live cell imaging of explant cultures of the intact intestine is not robust for our long-term imaging experiments at this point in time. In our own experience using the most recent protocols from the literature, intestinal explant cultures do not remain healthy for more than a few hours, and we do not wish to report results from less-than-healthy tissue. Therefore, we chose to do the experiment in the setting of the intact animal, where the continued health of the tissue is ensured. We appreciate the opportunity to include these data, as we think it is powerful to demonstrate that interspersion is relevant in vivo, and also that interspersion can be isolated from the chemical, mechanical and electrical cues present in vivo by using the minimal organoid system.

Reviewer #1:

[…] 1) The correlation between columnar epithelial height and the propensity to cause interspersion of daughter cells looks compelling but lack graphical analysis. It seems like a perfect place to include this would be in Figure 3C. Here, they could show a graphical correlation on cell height with wnt-treated organoids and the number of interspersed divisions or cells. It may be in the text but is not easy to see graphically and such a graph could make the point very clearly.

Thank you for this suggestion. We have now added these data as Figure 3D.

2) One thing that was not clear is whether all the cells within intestinal epithelia are interspersing or is it just the secretory lineage. This is important once we later find that the interspersion is linked to cell height. If so, we would expect all cells within the intestinal epithelium to intersperse. This would also predict that mitotic forces expected to drive cell migration up the villus will be dispersed, I expect.

Thank you for this suggestion. We have now included these data as Figure 1—figure supplement 1A. Please see point 2b in the response to the editorial summary, above.

3) Did the flatter visceral endoderm not intersperse in your videos? This would also lend more credence to the height as a controller.

Thank you for this suggestion. We have now included these data as Figure 3—figure supplement 1B. Please see point 2a in the response to the editorial summary, above.

Reviewer #2:

This manuscript examines how cell-types become interspersed during the development of intestinal organoids and presents evidence that this occurs as cells undergo cytokinesis at the apical surface and are then displaced from their siblings as they reintegrate towards the basal surface. This is an interesting question that takes advantage of the tractability of organoids for cell biological and live imaging approaches. While the manuscript is a useful contribution, I am not convinced that the data fully justify their conclusions. Most of the behaviours they observe, such as the apical movement and rounding of mitotic cells and planar spindle orientation have been well documented in a variety of other columnar or pseudostratified epithelia, so the main novelty lies in the mechanism of cell separation at cytokinesis. While their description is presumably accurate, it falls short of a mechanistic explanation and fails to consider the roles of the non-dividing basal neighbours or how cell junctions are remodelled during this process. As pointed out in the Discussion, divisions in the pseudostratified epithelia of the Drosophila imaginal discs share all of the features of the divisions in adult-derived intestinal organoids, including the polarised ingression of the cleavage furrow from basal to apical, yet the daughter cells are only very rarely separated during this process. Thus, any valid mechanistic model of cell dispersal in the intestine should be able to explain the Drosophila imaginal disc is different and the data as presented do not show any significant differences. The difference may lie in the behaviours of the neighbouring cells, as recent work has shown that the neighbours in the disc activate myosin contractility to pull on the apical junctions of the dividing cell during cytokinesis, which ensures that a new cell-cell junction forms between the newly-born daughters (Pinhiero et al., 2017).

We have now significantly rewritten the text and added new figures to address these important points (see also our response to point 1 in the editorial summary). Thanks to the reviewer’s comment, we realize that we over-emphasized cell aspect ratio and underplayed the central role that the geometry of cytokinesis plays in this process. In particular, a cornerstone of our paper is that a neighboring cell invades basally and laterally between the nascent daughters during cytokinesis. This can occur because the process of cytokinesis creates a minimal contact between the daughters, which allows an elongated neighboring cell to position between the daughters and displace them from one another. Thus, significant contact between daughter cells would be expected to prevent such intercalation. As the reviewer points out, extensive work in the literature has demonstrated that, during Drosophila cytokinesis, a long new adhesive contact forms between nascent daughters. This contrasts with the vertex contact between daughters in our system, and broadly observed during cytokinesis in vertebrate systems. The zippering of daughter cells together in the Drosophila system would be expected to represent a significant obstacle to invasion of a neighboring cell.

We realize, like any study, that this difference opens up questions for further experimentation. Probably the most relevant experiments would be to modify the degree of interaction between the daughter cells (decreasing it in Drosophila or increasing in the intestinal epithelium) as this would allow predictions of this model to be tested. However, we believe that strategies for manipulating cytokinesis in this manner are not entirely evident and overall this work is beyond the scope of this particular paper. However, we do now raise this possibility in the Discussion of this article.

We are grateful for the reviewer’s comment, which encouraged us to discuss the nature of the invasion mechanism in greater detail, helping us to hone in on the importance of cytokinesis geometry and minimal daughter-daughter contact in allowing interspersion to occur. We recognize that our discussion of the comparison with Drosophila in the previous draft was superficial and deserved more careful consideration, which we have now tried to do. We appreciate the opportunity to expand this discussion, which presented an opportunity to add more depth to our model. Indeed, the reviewer’s comment and the comparison with Drosophila has led us to re-craft the text and present our model more precisely.

I therefore think that these results, useful though they are, have not really solved the question of cell dispersal and the conclusions are over-interpreted. The results are a valuable contribution to the field, and this might be publishable if the conclusions were less over-blown. For example, they could concentrate on their observations that the apical-basal height of the epithelium is a key parameter for allowing dispersal.

Overall, in addition to addressing the important considerations of the Drosophila system as described above, we have now extensively revised the text to temper our conclusions and to focus throughout the paper on the mammalian systems for which we have direct data. As the reviewer suggests, we have focused the revised manuscript on our observations that the height of the epithelium is a key parameter for allowing dispersal, as well as incorporating discussion of the geometry of cytokinesis as described above.

The authors claim that cell-cell junctions on the apical surface promote the polarized, asymmetric mitotic cell-shape changes (Subsection “Cells intersperse during cytokinesis as part of a suite of cell shape changes restricted to the basolateral surface by cell-cell contact”/Figure 2B/Figure 2—figure supplement 1A-E). For this, they measure the positioning of cortical bleb formation in mitotic cells within the tissue context and after dissociation into single cells. They show that blebbing occurs only basally in the tissue (Figure 2—figure supplement 1A) and argue that this is inhibited on the apical side by cell-cell contacts ("…cell-cell junctions on the apical surface…oppose mitotic shape changes on the apical surface"), as blebbing occurs symmetrically in dissociated cells. However, in my opinion, the proposed mechanistic link between cell-cell junction and asymmetric cell shape changes are insufficiently addressed by means of these experiments. For instance, if cell-cell junctions on the apical side have an inhibitory effect on bleb formation, why is blebbing constrained to the basal surface (Figure 2—figure supplement 1A) and does not occur on the lateral sides? Is it possible that the "fine membranous processes that tether the cell to the basal surface" (subsection “Rearrangements of the actin cytoskeleton during cell division displace dividing cells along the apical-basal axis”, first paragraph and Figure 2—figure supplement 1G-H) have never detached from the basement membrane when the cell body moved apically.

We appreciate these points. As the blebbing is not a central component of the paper, we have now removed these data to avoid any confusion.

The dotted round cell outlines in Figures 2D, 2H, Figure 2—figure supplement 1L are confusing. It would be better to draw the outline of the cells based on the actual cell shape shown in the membrane signal.

Thank you for bringing this to our attention. We have now replaced these outlines with arrowheads, and show the membrane signal separately.

Figure 2B and Figure 2—figure supplement 1E: It is not clear why the authors suddenly switch to a cell line with MT +end marker when they analyze membrane blebbing in dissociated cells? Why not stick to the cell line with a membrane marker as in previous experiments?

We apologize for not including this information. EB3-GFP organoids were used to facilitate identification of organoids in mitosis after dissociation, when they cannot be identified by membrane morphology as all dissociated cells are spherical. We have now clarified this important point in the figure legends for Figure 2B (Figure 2—figure supplement 1E was removed, as described in the previous point). We thank the reviewer for catching this omission.

Reviewer #3:

The overarching question addressed in this study is that of how cells become interspersed within epithelia. This is a really interesting and universal open question with important implications. I was therefore really excited to get to read this paper, and unfortunately I need to admit I got a bit disappointed. The concluding statements made are bold, but the data presented, and more importantly system used for bulk of the analysis presented though quite clever, seems a little contrived. That said, this is not my field and I could easily be swayed.

The focus of this study is the mouse small intestine. Live (light-sheet and confocal etc.) imaging of small intestinal organoids in which subsets of cells are labelled in various genetic ways (inducible lineage-specific mosaic Cre lines, Confetti to mark clones etc.) is used to characterize the behavior of cells and begin to define mechanisms of cell interspersion within organoids as a proxy for the intestinal epithelium.

The authors use organoids derived from adult (tall) intestine and show that actin-, but not tubulin-, based behaviors drive cell interspersion, both cell rounding displacing cells apically at division and for subsequent re-establishment of a basal foothold. They then contrast these findings with fetal (short) intestinal-derived organoids which do not exhibit these behaviors (do they show this leads to a failure in interspersion and coherent clonal expansion within the epithelium?).

We show that organoids with short cells (fetal and grown in high Wnt) lead to a failure in interspersion and coherent clonal expansion. These data are included in Figure 3B and C and new Figure 3D.

The authors nicely describe how cell interspersion arises when interphase neighboring cells wedge between apically dividing daughters during cytokinesis, and requires an elongated (tall) epithelium, and that when the cellular aspect ratio is perturbed interspersion is disrupted resulting in local clonal expansion (and the production of clonal patches).

Even though separation of daughter cells has been described previously in various tissues in vivo, it is not clear whether the mechanisms defined here in organoids are at play within the crypts/villi, and no in vivo correlations with the in vivo situation are made for any of the behaviors described for organoids. I feel this is a major weakness of the study.

We appreciate this important point. As described in the editorial summary, point 3, above, we have now shown that interspersion of progeny also occurs in vivo. This is an important addition to the paper, and we thank the reviewer for bringing it to our attention.

https://doi.org/10.7554/eLife.36739.037

Article and author information

Author details

  1. Kara L McKinley

    1. Department of Cellular and Molecular Pharmacology, University of California, San Francisco, San Francisco, United States
    2. Howard Hughes Medical Institute, University of California, San Francisco, San Francisco, United States
    Contribution
    Conceptualization, Formal analysis, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing—original draft, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon 0000-0001-6283-9168
  2. Nico Stuurman

    1. Department of Cellular and Molecular Pharmacology, University of California, San Francisco, San Francisco, United States
    2. Howard Hughes Medical Institute, University of California, San Francisco, San Francisco, United States
    Contribution
    Formal analysis, Investigation, Methodology, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon 0000-0002-6179-8613
  3. Loic A Royer

    Chan Zuckerberg Biohub, San Francisco, United States
    Contribution
    Formal analysis, Methodology
    Competing interests
    No competing interests declared
  4. Christoph Schartner

    Department of Physiology, University of California, San Francisco, San Francisco, United States
    Contribution
    Investigation, Methodology
    Competing interests
    No competing interests declared
    ORCID icon 0000-0003-0599-3956
  5. David Castillo-Azofeifa

    1. Department of Orofacial Sciences, University of California, San Francisco, San Francisco, United States
    2. Program in Craniofacial Biology, University of California, San Francisco, San Francisco, United States
    Contribution
    Investigation, Visualization
    Competing interests
    No competing interests declared
  6. Markus Delling

    Department of Physiology, University of California, San Francisco, San Francisco, United States
    Contribution
    Investigation, Methodology
    Competing interests
    No competing interests declared
  7. Ophir D Klein

    1. Department of Orofacial Sciences, University of California, San Francisco, San Francisco, United States
    2. Program in Craniofacial Biology, University of California, San Francisco, San Francisco, United States
    3. Department of Pediatrics, University of California, San Francisco, San Francisco, United States
    4. Institute for Human Genetics, University of California, San Francisco, San Francisco, United States
    Contribution
    Resources, Supervision, Funding acquisition, Methodology, Writing—review and editing
    For correspondence
    Ophir.Klein@ucsf.edu
    Competing interests
    No competing interests declared
    ORCID icon 0000-0002-6254-7082
  8. Ronald D Vale

    1. Department of Cellular and Molecular Pharmacology, University of California, San Francisco, San Francisco, United States
    2. Howard Hughes Medical Institute, University of California, San Francisco, San Francisco, United States
    Contribution
    Conceptualization, Resources, Supervision, Funding acquisition, Methodology, Writing—review and editing
    For correspondence
    Ron.Vale@ucsf.edu
    Competing interests
    No competing interests declared
    ORCID icon 0000-0003-3460-2758

Funding

Howard Hughes Medical Institute

  • Kara L McKinley
  • Nico Stuurman
  • Ronald D Vale

National Institutes of Health (U01DK103147)

  • Kara L McKinley
  • David Castillo-Azofeifa
  • Ophir D Klein

Chan Zuckerberg Biohub

  • Loic A Royer

Damon Runyon Cancer Research Foundation (DRG-2282–17)

  • Kara L McKinley

Fritz Thyssen Stiftung

  • Christoph Schartner
  • Markus Delling

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

Acknowledgements

We thank Meghan Morrissey, Adam Williamson, Taylor Skokan, Amnon Sharir, Tom Wald, and other members of the Vale and Klein laboratories for reagents and helpful comments on the manuscript. We thank Dyche Mullins for feedback on the manuscript, Frederic de Sauvage (Genentech) for the Lgr5DTR-GFP allele, and Ilia Koev (Biogene) for his gift of a piece of Teflon AF 2400. Funding for this work was provided by the Howard Hughes Medical Institute (to RDV), the Thyssen Foundation (to MD), and the Chan Zuckerberg Biohub (to LAR). Research reported in this publication was also supported by National Institute of Diabetes and Digestive and Kidney Disorders (NIDDK) and National Institute of Allergy and Infectious Diseases (NIAID) of the National Institutes of Health under grant number U01DK103147 (to ODK). KLM is a Damon Runyon Fellow supported by the Damon Runyon Cancer Research Foundation (DRG-2282–17).

Ethics

Animal experimentation: All experiments involving mice were approved by the Institutional Animal Care and Use Committee of the University of California, San Francisco (protocol #AN151723).

Reviewing Editor

  1. Jody Rosenblatt, Reviewing Editor, University of Utah, United States

Publication history

  1. Received: March 17, 2018
  2. Accepted: June 8, 2018
  3. Accepted Manuscript published: June 13, 2018 (version 1)
  4. Version of Record published: June 28, 2018 (version 2)

Copyright

© 2018, McKinley 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

  • 2,608
    Page views
  • 398
    Downloads
  • 0
    Citations

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