|Home | About | Journals | Submit | Contact Us | Français|
The imaginal disc epithelia that give rise to the adult ectoderm of Drosophila can compensate to produce normal adult organs after damage. We looked at the local response to cell death using two genetic methods to elevate cell death rates. During cell competition, sporadic cell death occurs predictably along the boundaries between populations of competing wild-type and Minute/+ cells . Boundaries between Minute/+ and wild type populations showed an unusual degree of mixing, associated with mitotic re-orientation of wild type cells towards M/+ territory that they take over. Apoptosis of M/+ cells was the cue, and re-oriented mitosis required the planar cell polarity genes dachsous, fat, and atrophin genes. Clones mutated for pineapple eye, an essential gene, elevate apoptosis by a non-competitive mechanism . Mitosis was also re-oriented near cells mutant for pineapple eye, likewise dependent on the planar cell polarity genes. These findings show that planar cell polarity genes are required for responses to cell death. Oriented mitosis may help maintain morphology as dividing cells replace those that have been lost.
Many organisms can regulate their development in response to cell death or damage. Concentrated damage stimulates wound healing, but responses to the sporadic death of single cells have been less studied. During cell competition, sporadic cell death occurs predictably along the boundaries between populations of competing genotypes . The boundaries between competing genotypes also acquire unusually irregular courses, the basis for which has not been investigated. Cell migration is normally minimal in imaginal discs, and clones of cells derived from any marked precursor generally grow as coherent patches .
Unusual boundaries were first noticed when mitotic recombination was used to create clones of “Minute” cells (M/+), heterozygous for any one of many ribosomal protein genes , in predominantly +/+ compartments. Although M/+ flies are viable, M/+ clones in a wild type compartment exhibit high rates of cell death and are eliminated by this ‘cell competition’ [5, 6]. Prior to their elimination, M/+ clones get dispersed into small patches surrounded by +/+ cells . Conversely, even a small number of +/+ cells can progressively take over a M/+ developmental compartment . The boundaries of the +/+ clones were unusually jagged and irregular, indicating an unusual mixing between the cell populations as +/+ cells infiltrate formerly M/+ territory (Figure 1A–1G).
The anti-apoptotic baculovirus protein p35 was expressed to investigate whether cell death contributed to the irregular boundaries. Baculovirus p35 reduced the degree of intermingling between +/+ and M/+ cell populations (Figure 1H, 1I). Therefore, intermingling of +/+ cells with the remaining M/+ cells appears to be a response to death of individual M/+ cells.
It was surprising that cell death would create irregular boundaries between competing genotypes, because M/+ cells most surrounded by +/+ cells tend to die , which should smooth the boundaries by removing the most intermingled cells. We found that the number of mitotic figures was little affected by cell competition (Figure 2A), but that the plane of mitosis was re-oriented near competing boundaries. Because regional biases in mitotic orientation occur during normal development , mitotic orientation was measured with respect to boundaries between +/+ clones and M/+ cells rather than to global coordinates (Figure 2B). Since clone boundaries take many orientations, divisions of wild type cells show almost random orientation with respect to nearby clone boundaries, regardless of location in the disc (Figure 2C). By contrast, during cell competition the divisions of +/+ cells tended to be oriented perpendicular to nearby clone boundaries, directed towards the nearest M/+ cells (Figure 2D). Oriented cell division might be the cause of intermingling at boundaries between competing populations, and some examples were seen where dividing +/+ cells appeared to encroach on the M/+ territory (Figure 2E; Supplemental Figure 1). The mitotic orientation of M/+ cells on the other side of the boundary was much less affected (Figure 2F). Orientation of +/+ cells was lost when death of M/+ cells was prevented by co-expression of baculovirus p35 and dominant-negative Dronc (Figure 2G–H). Dominant-negative Dronc inhibits certain p35-independent aspects of apoptosis . Therefore, death of M/+ cells during cell competition orients the division of nearby +/+ cells to direct new-born +/+ cells into previously M/+ territory.
To test whether oriented cell division was specific for cell death caused by cell competition, we examined clones of cells mutant for the pineapple eye (pie) gene. The pie gene encodes a PHD-finger protein required for normal cell viability, so that clones of pie homozygous cells contain many apoptotic cells, but enough cells survive that pie homozygous clones are large enough for study (Figure 2I) . The apoptosis is not caused by cell competition since it does not require nearby +/+ cells. Mitotic +/+ cells were oriented by nearby clones of pie homozygous cells, showing that oriented cell division is not specific for competitive cell death, but a more general response to nearby apoptosis (Figure 2J). We also note that homozygous pie mutant clones had highly irregular outlines, like those of competing cell populations (Figure 2I).
A subset of the genes that are also important for planar cell polarity have recently been found to play a role in mitotic orientation during normal development, and to contribute to the normal morphology of the wing. Mitotic figures tend to orient towards the dorsal-ventral boundary in much of the wing portion of the wing imaginal disc, and cell clones are typically longer in the proximodistal axis, prefiguring the elongated proximodistal axis of the wing itself  (Figure 3A–C). This mitotic orientation is randomized in mutants of dachsous, a planar cell polarity gene that encodes an atypical cadherin also required as a tumor suppressor. Accordingly, dachsous mutant flies have a rounded wing shape. Mitotic orientation is also affected by over -expression of Fat, another atypical cadherin tumor suppressor required for planar cell polarity that is a heterophilic binding partner of Dachsous [8, 10–12]. We examined fat mutations, to confirm through loss-of-function studies that fat was required for mitotic orientation (Figure 3C,D). In contrast to the wild type, mitotic figures in clones of cells homozygous for a fat null mutation were un-oriented with respect to the dorsoventral boundary. Such clones were round, and their boundaries smooth (Figure 4Q, and data not shown). Thus fat is also required for mitotic orientation in the normal wing, along with dachsous.
The Fat intracellular domain interacts directly with the Atrophin protein, a transcriptional repressor that functions within the nucleus and is a homolog of Atrophin 1, in which expansion of a CAG domain is associated with human dentatorubral-pallidoluysian atrophy, an autosomal dominant neurodegenerative disease [13–16]. Mutations in atrophin are embryonic lethal, but clones of null mutant cells exhibit planar polarity defects and grow with smooth boundaries, similar to dachsous or fat mutants . We examined mitotic figures in atro mutant clones to determine whether atro was also required for mitotic orientation. We found that atro mutant cells showed no particular orientation with respect to the proximodistal axis (Figure 3G–H). The atro mutant clones also had smooth boundaries (Figure 3E–F). Thus the atrophin gene appears to function in the same pathway as fat and dachsous in orienting mitosis in normal development.
Smooth interfaces between cell populations can indicate affinity differences between cells . To test the effect of cell adhesion, clones of cells that overexpressed E-cadherin were examined. These have smooth boundaries, like fat, dachsous or atro mutant clones. The orientation of mitosis with respect to the proximo-distal axis was not affected by overexpression of E-cadherin, however (Figure 3I,J). We suggest that whereas cells over -expressing E-cadherin form smooth boundaries due to cell affinity differences , fat, dachsous or atro mutations do so through random mitotic orientation, their clones growing like colonies in the tissue. We cannot exclude that cell adhesion differences might also contribute, given heterophilic binding between Fat and Dachsous , and because Atro might affect transcription of cell adhesion molecules.
To determine whether mitotic orientation in response to cell death also depends on the same PCP genes, clones of homozygous dachsous, fat, or atrophin cells were examined in M/+ backgrounds, and their mitotic orientations measured with respect to the clone boundaries. Unlike clones of +/+ cells, mitosis of ds mutant, ft mutant, or atro mutant cells was not oriented by the boundary with M/+ cells (Figure 4A–F). The boundaries of the mutant clones remained smooth in the M/+ background (Figure 4G–J and data not shown). Thus these PCP genes were required to re-orient cell division in response to competitive cell death, as well as for oriented cell division in normal development.
The failure of dachsous, fat and atro mutant cells to orient in response to M/+ cells could be explained in a trivial way if they did not compete with M/+ cells, since we have already noted that mitotic orientation depended on death of M/+ cells (Figure 2G–H). This was plausible because competitive death of M/+ cells is increased by the proximity and degree of their exposure to non -Minute cells ; such exposure may be reduced by the smooth boundaries with fat and atro clones. Indeed compartment boundaries, across which no intermingling occurs, are also barriers to competition [5, 19, 20].
Potential effects of dachsous or fat mutations on cell competition were hard to assess, because these mutant cells are themselves hyperplastic . The atro mutations were not hyperplastic, however. Clones of cells mutant for atro grew at a rate similar to controls, and did not cause cell death (Figures 3E, ,4K,4K, and data not shown). This permitted measurement of the effect of atro mutations on cell competition. If atro mutant cells did not compete with M/+ cells, we would expect atro/atro clones to grow less in a M/+ background than wild type cells do, and not to kill neighboring M/+ cells as wild type cells do. We found, however, that atro/+, M/+ cells were out-competed by atro/atro clones almost as well as by wild-type clones, despite the smooth boundaries between them (Figure 4G–J). Clones of atro mutant cells grew rapidly in a M/+ background, occupying large proportions of the tissue, like clones of +/+ cells. The atro/atro clones were slightly smaller in the Minute background, but the difference was not statistically significant (Figure 4K,L). Like M/+ cells next to +/+ cells, atro/+, M/+ cells died next to atro/atro cells (data not shown). The rate of death was reduced in absolute terms compared to M/+ cells next to +/+ cells, but similar when adjusted for the shorter boundaries of the smooth atro clones (Figure 4 M,N). These experiments confirmed that atro/atro cells competed with and killed M/+ cells, and therefore that atro was required for re-orienting cell division in response to M/+ cell death. We could not assess the consequences of failing to re-orient cell divisions for wing shape because of the additional roles of atro in adult wing differentiation . We also found that M/+ cells continued to die next to clones of +/+ cells given smooth boundaries through overexpression of E-cadherin. There was death within E-cadherin overexpression clones too (data not shown), presumably because of dominant-negative effects of E-cadherin overexpression on Wg survival signaling.
We also assessed whether PCP genes were required to reorient mitosis in response to the non-competitive death of pie/pie cells. Mitotic recombination was induced in the ft +/+ pie genotype to obtain ft/ft clones next to pie/pie clones. The ft/ft cells adjacent to pie/pie clones showed random mitotic orientation, whereas the ft +/+ pie cells that were near to pie/pie clones did re-orient (Figure 4O,P). Whereas the boundaries between pie/pie clones and ft +/+ pie cells were irregular, and boundaries between ft/ft clones and ft +/+ pie cells were smooth, boundaries between pie/pie clones and ft/ft clones were smooth (Figure 4Q). Because cell death still occurred in pie/pie clones near to ft/ft cells (data not shown), these mitotic orientation data provide further evidence that PCP genes are required to reorient mitosis in response to cell death.
Because competitive cell death is compensated by growth , and Fat can affect the Hippo-Salvador-Warts (HSW) pathway of tumor suppressors , we wondered whether HSW activities might be altered where +/+ and M/+ cells abut, but were unable to detect changes in the expression of the Expanded or Merlin proteins, of the expanded LacZ enhancer trap, or of the nuclear localization of the Yorkie protein where +/+ and M/+ cells are adjacent (data not shown); these are all reporters of HSW activity .
Our main conclusion is that cell death orients nearby cell division in wing disc development. We suggest that regulative development to replace dead cells requires cell rearrangement to reconstitute organ shape in addition to cell number regulation, and that in imaginal discs this occurs by mitotic re-orientation. Since each organ is the sum of its constituent clones of cells, behavior that affects the shapes of clones ultimately defines the shape of each organ. The clone shapes reveal the ancestry of individual cells over development, and irregular boundaries reflect the preferential origin of cells from one cell population.
Re-orientation in response to cell death requires the Fat/Dachsous PCP pathway. This pathway also affects cell division axis in normal wing development, although very little cell death occurs in normal development [20, 24]. Developmental signals and cell-death derived signals must converge on mitotic orientation at some level upstream of PCP genes. It is not known at present how cells sense nearby apoptosis. Possible mechanisms include alterations in Fat or Ds activity levels in dying cells, which would have non -autonomous effects , expression of growth factors by dying cells , cell polarization in the course of engulfing dead cell fragments , or mechanical stretching of the epithelium when cell numbers are reduced . PCP proteins could sense one or more such signal and define the polarized cellular response; alternatively, PCP genes might be necessary to implement a polarity that is sensed by other proteins.
The mitotic orientation role of PCP genes in development may be conserved: the mammalian Fat-4 gene is required for oriented cell division in kidney tubule elongation ; polarized migration of smooth muscle cells following arterial damage or atherosclerosis depends on mammalian Fat-1 and Atrophins [28, 29]. It will interesting to discover whether planar cell polarity genes are involved in the response to damage and cell death in mammals, and whether defects in such processes contribute to conditions such as dentatorubral-pallidoluysian atrophy. It seems plausible that cell migration or convergent extension could regulate in response to damage also, in tissues where such processes are more prominent [30–32].
Mosaic clones were obtained using the FRT-FLP method for mitotic recombination . Flies were maintained at 25°C. Heat shock was usually performed at 37°C for 1 hour. Minute genotypes were generally heat shocked 84±12 h after egg laying and dissected 72 h later, unless indicated otherwise. Other genotypes were heat shocked at 60±12 h after egg laying and dissected 60 h later. Except for Figure 1H, the M/+ genotypes shown in Figure 1 and Figure 4A and 4C were heat shocked for 30min at 108±12 hrs AEL and dissected 48 hrs later.
As the difference in clone shapes was not captured well by the circularity statistic (area/perimeter), we used a zero-crossing statistic we called the ‘turn index’ to describe the irregularity of clone boundaries. The turn index was defined by the number of turns greater than 90 degrees leftwards encountered during a clockwise circuit of the clone perimeter, normalized against the length of the clone perimeter (see Figure 1G).
Most imaginal disc cells divide symmetrically in the plane of the epithelium. The axis of division was measured in anaphase and telophase cells labelled by antibodies for phospho-histone 3, a marker for mitotic chromatin. We found this superior to spindle components such as gamma-tubulin, because of both the extensive tubulin labelling of these epithelia and the need to define cell cycle stage given the potential for the spindle to rotate early in mitosis. Accuracy in distinguishing anaphase and telophase nuclei from pairs of nearby cells in prophase was validated by labeling some samples for the cortical protein Merlin to outline cell boundaries. The dorsoventral boundary was positioned using antibodies against the Senseless protein which is expressed in neurons along the future wing margin . To quantify orientation with respect to clone boundaries, the boundary angle was defined as follows: Projecting the division axis towards the Minute cell population, the boundary was defined as the straight line connecting two points approximately one spindle-length apart in each direction from the intersection, along the interface between the Minute and non-Minute populations, as shown in Figure 2B. Following this procedure provided objectivity in cases where the boundary between cell populations was irregular. Only mitotic figures close to the clone boundary were assessed, as mitoses in the clone interior may be equidistant to boundaries in several directions.
see supplemental methods online
Supplemental Figure 1
The confocal data showing an oriented mitosis impinging on the territory of neighboring M/+ cells, shown from an apical viewpoint in Figure 2E, is shown here resliced to give a side view. The apical wing disc surface is uppermost. The mitotic figure is deforming the M/+ cells within the plane of the epithelium, not flattened above them apically.
We thank H. McNeill, P. Meier, J.P. Vincent, and the National Stock Center at Bloomington Indiana for Drosophila strains, and R. Fehon and the Developmental Studies Hybridoma Bank for antibodies. The manuscript was improved by comments from A. Jenny, L. Johnston, H. McNeill, N. Sibinga and D. Tyler. Data in this paper are from a thesis to be submitted in partial fulfillment of the requirement for the degree of Doctor of Philosophy in the Graduate Division of Biomedical Sciences, Albert Einstein College of Medicine, Yeshiva University, USA. Supported by a grant from the NIH (GM61230).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.