|Home | About | Journals | Submit | Contact Us | Français|
Mutations that inactivate either merlin (mer) or expanded (ex) result in increased cell growth and proliferation in Drosophila. Both Mer and Ex are members of the Band 4.1 protein superfamily, and, based on analyses of mer ex double mutants, they are proposed to function together in at least a partially redundant manner upstream of the Hippo (Hpo) and Warts (Wts) proteins to regulate cell growth and division. By individually analyzing ex and mer mutant phenotypes, we have found important qualitative and quantitative differences in the ways Mer and Ex function to regulate cell proliferation and cell survival. Though both mer and ex restrict cell and tissue growth, ex clones exhibit delayed cell cycle exit in the developing eye, while mer clones do not. Conversely, loss of mer substantially compromises normal developmental apoptosis in the pupal retina, while loss of ex has only mild effects. Finally, ex has a role in regulating Wingless protein levels in the eye that is not obviously shared by either mer or hpo. Taken together, our data suggest that Mer and Ex differentially regulate multiple downstream pathways.
During the normal development of an organism and in disease states such as cancer, changes in cell shape and cell-cell adhesion often correlate with changes in cell proliferation. Many members of the protein 4.1 superfamily link the cortical actin cytoskeleton to membrane proteins including receptors for growth factors [reviewed in (Bretscher et al., 2002; Sun et al., 2002)] and are uniquely positioned to modulate cell proliferation in response to alterations in cell morphology. Protein 4.1 family members share a conserved FERM (Four-point-one, Ezrin, Radixin, Moesin) domain, which is typically N-terminal, and many also contain a spectrin-actin binding domain (Bretscher et al., 2002; Sun et al., 2002). In mammals, it appears that at least two FERM domain-containing proteins, Merlin and DAL-1, function as tumor suppressors (Rouleau et al., 1993; Tran et al., 1999; Trofatter et al., 1993).
Two FERM domain-containing proteins that negatively regulate growth and proliferation in Drosophila are Merlin (Mer) and Expanded (Ex) (Boedigheimer and Laughon, 1993; LaJeunesse et al., 1998). Mer is the functional homolog of human Merlin, which is encoded by the NF2 gene. NF2 is a classical human tumor suppressor gene that is mutated in the tumor syndrome neurofibromatosis type 2, a disease characterized by benign tumors of the central nervous system, particularly schwannomas and meningiomas (Rouleau et al., 1993; Trofatter et al., 1993). Mer is most closely related to Ezrin, Radixin, and Moesin (the ERM proteins), which, via their C-termini, link the cytoskeleton to membrane proteins (Bretscher et al., 2002; Xiao et al., 2003). Though Mer does not contain a C-terminal spectrin-actin binding domain, it may bind actin via its FERM domain (Brault et al., 2001; Xu and Gutmann, 1998). Loss of Mer function in mammalian cell culture appears to destabilize adherens junctions and results in loss of contact-dependent inhibition of proliferation (Lallemand et al., 2003; McClatchey and Giovannini, 2005; Okada et al., 2005).
Drosophila Ex protein is structurally distinct from Mer and is phylogenetically distinct from the ERMs (Boedigheimer and Laughon, 1993; Bretscher et al., 2002). Ex contains several potential protein-protein interaction domains in its C-terminus that are not present in Mer, including three putative Src Homology 3 (SH3)-binding domains, several homopolymeric amino acid stretches, and multiple QA and LX repeats (Boedigheimer and Laughon, 1993). It is not known whether Ex binds to actin, and there is currently no unequivocal evidence that a functional homolog of Ex exists in mammals.
Although Merlin and Expanded are dissimilar FERM domain-containing proteins, their functions have been linked in several ways. Early studies showed that the Mer and Ex proteins co-localize with cortical actin in the apical region of the cell (Boedigheimer and Laughon, 1993; Boedigheimer et al., 1997). Additional work uncovered genetic and physical interactions between Mer and Ex: loss of one copy of mer dominantly enhanced wing overgrowth in an ex mutant, and fragments of Mer and Ex protein are capable of interacting physically with each other in far-Western experiments or when overexpressed in cultured cells (McCartney et al., 2000).
Recent work from two groups has shown that clones doubly mutant for mer and ex have more profound phenotypes than either single mutant (Hamaratoglu et al., 2006; Maitra et al., 2006). For this reason, these previous studies have focused primarily on the phenotypes of mer ex double mutant clones. One study showed that loss of mer and ex together resulted in defective endocytic trafficking, leading to the upregulation of multiple cell growth and proliferation pathways, including the Epidermal Growth Factor Receptor (EGFR) and Notch pathways (Maitra et al., 2006). The phenotypic similarities between mer ex mutant clones and clones mutant for components of the Hippo (Hpo)/Salvador (Sav)/Warts (Wts) growth pathway have also been described (Hamaratoglu et al., 2006).
Hpo and Wts are protein kinases of the Ste20 and the nuclear Dbf-2 related (NDR) families, respectively (Harvey et al., 2003; Jia et al., 2003; Justice et al., 1995; Pantalacci et al., 2003; Udan et al., 2003; Wu et al., 2003; Xu et al., 1995). Hpo likely activates Wts by phosphorylating it, and Sav and Mats (Lai et al., 2005) facilitate the activation of Wts. An important downstream target of the Hpo/Sav/Wts pathway is Yorkie, a transcriptional co-activator that is negatively regulated by Wts (Huang et al., 2005). A reduction in the activity of the Hpo/Sav/Wts pathway results in increased growth, delayed cell cycle exit, ectopic cell survival, and upregulation of downstream target genes, including thread, which codes for Drosophila Inhibitor of Apoptosis Protein 1 (DIAP1), cyclin E, ex, and mer (Hamaratoglu et al., 2006; Harvey et al., 2003; Jia et al., 2003; Kango-Singh et al., 2002; Pantalacci et al., 2003; Tapon et al., 2002; Udan et al., 2003; Wu et al., 2003).
The finding that the mer ex double mutants have more severe phenotypes than the single mutants implies that the two genes may function in a redundant manner to regulate the same downstream pathways or may function in distinct pathways affecting growth and differentiation. It has been proposed that ex and mer function together, possibly cooperatively, to restrict tissue growth (Hamaratoglu et al., 2006; Maitra et al., 2006). Combined overexpression of Merlin and Expanded has been shown to be more effective than overexpression of either protein alone in repressing Yorkie-dependent transcription in tissue culture cells (Hamaratoglu et al., 2006). However, it is unlikely that the two proteins function together solely in a stoichiometric complex where both proteins are required for the activity of the complex. If that were the case, then inactivation of either mer or ex alone should have a result that is similar to the inactivation of both genes. Thus the mechanisms by which the functions of the two proteins are related are still unresolved.
To understand how inactivation of mer and ex cause tissue overgrowth, we have examined the phenotypic consequences of disrupting either mer or ex alone and show that they perform distinct functions. Mer and Ex are each capable of restricting cell growth and cell cycle progression. However, Ex has a clear role in regulating cell cycle exit, while Mer regulates apoptosis. Thus, the mer ex double mutant phenotype is a synergy of the delayed cell cycle exit phenotype of ex and the ectopic cell survival phenotype of mer. Additionally, Ex has a function in regulating Wingless protein levels in the developing eye that is not obviously shared by Mer. Our data suggest that Ex and Mer regulate growth, proliferation, and apoptosis in ways that are both qualitatively and quantitatively different from each other.
Stocks listed below were constructed for this study or have been described as indicated. The following alleles and stocks were used to construct some stocks below: w; GMR-p35 (III) (Hay et al., 1994), wg1 (Sharma, 1973), eyFLP (II) (Newsome et al., 2000), w; exe1/CyO (Boedigheimer et al., 1993).
w1118 sn3 FRT19A (Xu and Rubin, 1993), w1118 sn3 merB2 FRT19A/FM7a, w1118 sn3 merF705 FRT19A/FM7a, y w Ubi-GFP FRT19A; eyFLP (II) (Moon et al., 2005), y w P[m-w+ arm-lacZ] FRT19A; eyFLP/TM3 (gift of J. Treisman), y w l(1)cl 8.7 P[m-w+ arm-lacZ] FRT19A/FM7a; eyFLP (II), y w mer4 FRT19A/FM7a Act-GFP (Fehon et al., 1997), y w mer4 FRT19A; P[cos mer+]/TM3 (Fehon et al., 1997), w1118 sn3 FRT19A; GMR-p35(III), w1118 sn3 merB2 FRT19A/FM7a; GMR-p35(III).
w; FRT40A (Xu and Rubin, 1993), w; exMGH1 FRT40A/CyO, w; exMGH2 FRT40A/CyO, w; exe1 FRT40A/CyO, w; wg1 FRT40A/CyO, w; exMGH1 wg1 FRT40A/CyO, w; FRT40A; GMR-p35(III), w; exMGH1 FRT40A/CyO; GMR-p35(III), y w eyFLP; P[m-w+ arm-lacZ] FRT40A (Delalle et al., 2005), y w eyFLP; Ubi-GFP FRT40A (Delalle et al., 2005); y w hsFLP; Ubi-GFP FRT40A (Delalle et al., 2005), y w eyFLP GMR-lacZ; l(2)cl-L31 P[w+] 30C FRT40A/CyO y+ (Newsome et al., 2000).
Mutant alleles were generated by EMS mutagenesis of w; FRT40A (for ex alleles) or w1118 sn3 FRT19A (for mer alleles) and identified by their ability to cause overgrowth of mutant tissue relative to wild type tissue in mosaic eyes during eye development. exMGH1 is a 747 base pair deletion mutation that shifts the ex reading frame after amino acid 1058 and truncates the protein 49 amino acids after the frame shift. exMGH2 is an eight base pair deletion mutation that shifts the ex reading frame after amino acid 957 and truncates the protein 34 amino acids after the frame shift. In our hands, exMGH1 behaves similarly to exe1 in all ways tested: both exMGH1 and exe1 tissue have increased DIAP1, Cyclin E, and Cyclin B levels. In addition, both alleles have very similar phenotypes in the mature pupal retina, in BrdU assays in larval eye discs, and when flipped over a cell lethal mutation.
merB2, which is identical to the mutation present in mer4 (Fehon et al., 1997), is a C to T base change that changes the CAG Glu codon at position 170 to a TAG stop codon. merF705 is a two base pair insertion that shifts the mer reading frame at amino acid 202 and truncates the protein two amino acids after the frame shift.
l(1)cl 8.7 is a cell lethal mutation on the X chromosome isolated in a FLP/FRT growth and proliferation screen. Tissue homozygous for l(1)cl 8.7 is highly underrepresented in mosaic adult eyes when compared to wild type controls. Additionally, l(1)cl 8.7 mosaic eyes appear to be normally patterned when viewed under the dissecting microscope.
Antibodies and dilutions used: mouse anti-Wg 1:100 (4F4, DSHB); rat anti-ELAV 1:200 (7E8A10, DSHB); guinea pig anti-Cyclin E 1:1000 (Terry Orr-Weaver); mouse anti-Cyclin A 1:5 (A12, DSHB); mouse anti-Cyclin B 1:5 (F2F4, DSHB); mouse anti-DIAP1 1:200 (Bruce Hay); rabbit anti-Ex 1:500 (Allen Laughon); mouse anti-Dlg 1:50 (4F3, DSHB); mouse anti-BrdU 1:100 (Becton Dickinson); rabbit anti-cleaved caspase 3 1:200 (Cell Signaling); Secondary antibodies (Jackson Immuno Research, Molecular Probes) were used at dilutions between 1:200 and 1:1000.
Imaginal discs were dissected in room temperature (RT) Schneider’s medium containing 10% fetal bovine serum. Discs were held in RT Schneider’s for up to 10′ and transferred to 1.5 ml tubes containing 500 μl RT Schneider’s containing 10mM BrdU. The tubes were incubated for 30′ or 90′ at RT in a rotating tube inversion shaker. Discs were washed 1X in RT Schneider’s and 2X in RT phosphate-buffered saline (PBS), pH 7.4. Discs were fixed for 16–20 hours at 4°C in 1.5% formaldehyde/0.01% Tween 20 in PBS. Discs were then washed 5X with PBS and DNAse treated for 45′ at 37°C [5 μl RQ1 DNAse (Promega), 10 μl DNAse reaction buffer, and 85 μl water]. Discs were washed 3X with PBS/0.3% Triton X-100 and stained for two hours at RT with anti-BrdU (1:100) in 10% normal goat serum and PBS/0.1% Triton X-100. Discs were then washed 5X in PBS/0.3% Triton X-100 and incubated for two hours at RT with donkey-anti-mouse rhodamine red X (1:400) in 10% normal goat serum and PBS/0.1% Triton X-100. Discs were then washed 5X in PBS/0.3% Triton X-100 and transferred into mounting solution.
The numbers of cells per ommatidial unit were counted following the method described by Wolff and Ready (Wolff and Ready, 1991). All cells within a hexagonal target area defined by an ommatidium and its six nearest neighbors were counted to determine the number of cells per ommatidial unit/target area. Cells bisected by the target area boundary were counted as one half a cell. A typical wild type target area contains 21 cells: four cone cells, two primary pigment cells, six secondary pigment cells, three tertiary pigment cells, three bristle cells, and six secondary pigment cells which are bisected by the target area boundary (three total cells). Counting data in the text is presented in the format cell count ± standard deviation. P-values were generated using unpaired Student’s t-tests.
We identified two alleles of expanded (ex) and two alleles of merlin (mer) in genetic screens for mutations that increase the ratio of mutant tissue to wild type tissue in mosaic adult eyes. exMGH1 and exMGH2 are lethal alleles that truncate the C-terminus of Ex and have a strong overproliferation phenotype (Fig. 1A, B, and data not shown). Both exMGH1 and exMGH2 produce truncated Ex protein at high levels as detected by an antibody raised against the Ex N-terminus (see below). However, in all ways we have tested (see Materials and Methods), exMGH1 behaves similarly to exe1, a strong allele resulting from the excision of a 5′ P-element (Boedigheimer and Laughon, 1993). The deletion associated with exe1 removes the first exon of ex (a non-coding exon), and exe1 tissue in the eye has strongly reduced Ex protein levels (data not shown). The phenotypic similarity between these molecularly distinct alleles of ex indicates that the Ex C-terminus is required for the growth suppressive function of the protein.
merB2 and merF705 are lethal alleles that truncate the C-terminal two-thirds of the Mer protein. Both merB2 and merF705 result in a proliferative advantage indistinguishable from mer4 (Fehon et al., 1997), a null allele of mer (Fig. 1C, D, and data not shown), and behave as genetic nulls based on our phenotypic and molecular analyses.
Both mer and ex mosaic eyes are overgrown relative to control mosaic eyes. Adult eyes mosaic for mer contain ~50 more ommatidia than controls and are larger than control eyes in both the anterior-posterior (AP) and the dorso-ventral (DV) dimensions (Fig. 1G, H). mer mosaic eyes contain approximately one more ommatidium per vertical column and one more column of ommatidia per eye than wild type controls (data not shown). ex mosaic eyes can also be significantly larger than wild type controls (see below for exceptions), and overgrown ex mosaic eyes contain ~75 more ommatidia than wild type controls (Fig. 1E, F). Overgrown ex mosaic eyes appear to be larger than controls in both the DV and AP dimensions (Fig 1E, F), though patterning irregularities prevent definitive quantification of the number of columns per eye and the number of ommatidia per column.
There are several qualitative differences in the phenotypes of adult eyes mosaic for ex or mer. Despite being broader than control eyes, mer mosaic eyes (Fig. 1H) are similar in shape and patterning to control mosaic eyes (Fig. 1G). In contrast, ex mosaic eyes (Fig. 1F) are slightly rough, have ectopic bristles (Fig. 1F, inset), and are irregularly shaped when compared to control eyes (Fig. 1E). Consistent with a previous study (LaJeunesse et al., 1998), we found that mer mutant ommatidia (Fig. 1K) in mosaic adult eyes were patterned normally (Fig. 1I). In contrast, ex mutant ommatidia frequently show defects in both ommatidial chirality and rotation (Fig. 1J, blue and red arrows), as previously reported (Blaumueller and Mlodzik, 2000). Importantly, we also observed non-autonomous ommatidial chirality and rotation defects; some ommatidia comprised entirely of wild type photoreceptors located near clonal boundaries were misrotated (data not shown), while some had adopted the mirror-image chirality of those located on the opposite side of the equator (Fig. 1J, black arrow) Since ommatidial chirality and rotation is determined by the R3 and R4 photoreceptor cells (Tomlinson and Struhl, 1999; Zheng et al., 1995), our data suggest that loss of ex affects retinal patterning in both the mutant and the adjacent wild-type tissue.
The eye, wing, and antennal imaginal discs from third instar exMGH1/exMGH2 larvae (Fig. 2B, F) were greatly overgrown compared to wild type imaginal discs (Fig. 2A, E), consistent with previous studies (Blaumueller and Mlodzik, 2000; Boedigheimer and Laughon, 1993). Despite being overgrown, eye discs from ex transheterozygotes had greatly reduced numbers of differentiated photoreceptors (Fig. 2B). Furthermore, ~10% of ex homozygous eye discs had no differentiated photoreceptors and were much smaller than their overgrown antennal discs (data not shown). Thus, ex imaginal disc tissue can overgrow, and ex mutant cells in the eye are capable of differentiating into photoreceptors in ex mosaics, but complete loss of ex inhibits eye differentiation, resulting in small eyes in ex homozygotes. These data suggest that wild type tissue is necessary for ex mosaic eye tissue to differentiate properly and produce an overgrown adult eye.
To investigate the role of wild type tissue in ex mosaic eye phenotypes further, we created mosaic eyes in which the homozygous wild type tissue is almost completely eliminated by the presence of a cell lethal mutation (Newsome et al., 2000). Mosaic eyes composed mostly of ex tissue (Fig. 2J) were much smaller than control eyes (Fig. 2I) and were similar in size to ex homozygous eyes in pharate adults (data not shown). This confirms a role for wild type tissue in photoreceptor differentiation and eye overgrowth in ex mosaics.
In contrast to ex, third instar larval eye discs, antennal discs, and wing discs of mer hemizygotes (Fig. 2D, H) were very similar in size to those from control larvae hemizygous for the parent chromosome (Fig. 2C, G). Loss of mer function results in a slight increase in eye disc size (Fig. 2D), but does not result in the dramatic tissue overgrowth phenotypes seen with loss of ex (Fig. 2B, F). In addition, the eye discs of mer hemizygotes contained a normal complement of differentiated photoreceptors (Fig. 2C, D), unlike what is seen with ex (Fig. 2B).
Specification of photoreceptor clusters in the developing eye occurs as the morphogenetic furrow moves across the eye disc (Ready et al., 1976; Tomlinson and Ready, 1987; Wolff and Ready, 1993). Morphogenetic furrow progression is negatively regulated by Wingless (Wg), a secreted glycoprotein normally produced at the dorsal and ventral margins of the eye imaginal disc (Fig. 3A) (Baker, 1988; Ma and Moses, 1995; Treisman and Rubin, 1995). Wg restricts morphogenetic furrow progression by antagonizing Dpp, which promotes morphogenetic furrow progression (Chanut and Heberlein, 1997; Heberlein et al., 1993; Ma and Moses, 1995; Pignoni and Zipursky, 1997; Treisman and Rubin, 1995). Because delayed morphogenetic furrow progression and non-autonomous defects in ommatidial chirality and rotation are consequences of increased Wg signaling (Kumar and Moses, 2001; Lee and Treisman, 2001; Tomlinson et al., 1997; Treisman and Rubin, 1995; Wehrli and Tomlinson, 1998), we investigated whether Wg protein levels were elevated in ex mutant tissue.
Wg protein is detectable at the dorsal and ventral margins of control third instar larval eye discs (Fig. 3A). ex mosaic eye discs with large clones overlapping the dorsal or ventral margins of the eye disc contain both higher levels of Wg protein and enlarged domains of Wg protein within clones (Fig. 3B). This finding suggested that ectopic Wg protein in ex clones might be responsible for preventing morphogenetic furrow progression, as eye discs with ectopic Wg protein frequently had a delayed morphogenetic furrow and were missing differentiated photoreceptors (Fig. 3C). Consistent with this observation, ~25 % (135/538) of ex mosaic eyes were smaller than wild type mosaic controls because a significant portion of the adult eye (either dorsal or ventral) has been transformed into cuticle or bristle (data not shown).
In contrast to the delayed morphogenetic furrow, elevated Wg protein levels, and enlarged domain of Wg protein that we detected in ex clones, we did not detect defects in furrow progression, increased Wg protein levels or ectopic domains of Wg production in mer tissue. Wg protein staining in mer mosaic eye imaginal discs (Fig. 4B) was very similar to that seen in control mosaic discs (Fig. 4A). Our data shows that ex regulates Wg protein levels, while mer does not detectably do so.
To determine whether reducing Wg levels in ex tissue suppresses the small eye phenotype associated with eyes composed only of ex tissue, we constructed a recombinant chromosome containing exMGH1 and wg1, a hypomorphic allele of wg. wg1 contains a regulatory mutation outside of the wg coding sequence (Neumann and Cohen, 1996; Sharma, 1973; Sharma and Chopra, 1976; van den Heuvel et al., 1993). The presence of the wg1 allele on the exMGH1 mutant chromosome partially suppressed the small eye phenotype of eyes composed almost entirely of ex mutant tissue (compare Fig. 3F to Fig. 2J) and reduced by three-fold the proportion of ex mosaic eyes that were smaller than controls because of transformation to a portion of the eye field to bristle or cuticle (46/588, ~8%). In addition, exMGH1 wg1 mosaic eye discs had Wg protein staining patterns that more closely resembled wild type mosaics than exMGH1 mosaic eye discs (Fig. 3D). Creating eyes mosaic for wg1 alone did not appear to substantially affect the ratio of mutant to wild type tissue, eye size, or external morphology of the eye (Fig. 3G).
Because Wg is thought to function as a mitogen in certain contexts in the developing wing (Giraldez and Cohen, 2003; Johnston and Sanders, 2003), we were interested to know whether Wg upregulation contributes to the ex overproliferation phenotype in the eye. Wg protein appears to contribute to the large eye phenotype of ex mosaic eyes (Fig. 1A, B), since exMGH1 wg1 mosaic eyes (Fig. 3H) are similar in size to wg1 mosaic eyes (Fig. 3G) and smaller than exMGH1 mosaic eyes (Fig. 1B). However, reducing Wg function does not noticeably change the ratio of mutant to wild type tissue in ex mosaic eyes (compare Fig. 1B and Fig. 3H), suggesting that increased Wg protein does not alter the intrinsic ability of ex tissue to overproliferate relative to wild type tissue.
Our results indicate that increased Wg protein is at least partly responsible for inhibiting morphogenetic furrow progression in ex mutant tissue in the eye. Increased Wg protein in ex tissue explains why ~25% of ex mosaic eyes are smaller than wild type controls and why eye imaginal discs composed solely of ex tissue fail to differentiate into photoreceptors. Though increased Wg protein does not appear to contribute to the proliferative advantage of ex cells, it may contribute to the eye overgrowth phenotype of ex mosaic eyes by slowing morphogenetic furrow progression and allowing cells in the anterior portion of the disc to proliferate for a longer period (see Discussion).
Since ex and mer are proposed to negatively regulate growth by activating the Hpo/Sav/Wts pathway (Hamaratoglu et al., 2006), we also investigated whether loss of Hpo, which results in significant tissue overgrowth (Fig. 4E, F), also resulted in Wg deregulation in the eye. We tested this by examining Wg protein levels in hpo eye tissue and by creating eyes composed almost exclusively of hpo tissue. Eye discs mosaic for hpo42–47, a null allele of hpo (Wu et al., 2003), had normal levels and domains of Wg protein (Fig. 4C, D). Eyes composed almost entirely of hpo tissue (Fig. 4H) were overgrown and noticeably larger than both control (Fig. 4G) and hpo mosaic eyes created using a viable tester chromosome (Fig. 4F). Since Wg protein levels appear normal in hpo mosaic eye discs, Wg regulation by ex in the eye may occur independently of Hpo.
To determine whether mer or ex regulates cell cycle exit, we investigated whether loss of ex or mer resulted in ectopic S-phase entry in normally postmitotic cells in larval eye discs. We consistently detected clear ectopic BrdU incorporation posterior to the second mitotic wave in ex clones in larval eye discs (Fig. 5A). In contrast, we did not detect ectopic S-phase entry in mer clones posterior to the second mitotic wave (Fig. 5B). Because this important distinction between the mer and ex mutant phenotypes is not in agreement with a previous study that reported BrdU incorporation posterior to the second mitotic wave in third instar larval eye discs from mer homozygotes (Maitra et al., 2006), we repeated our experiment several times using short (30′) and long (90′) BrdU pulses and multiple ex and mer alleles. With a 90′ BrdU pulse, 18/19 ex mosaic eye discs exhibited ectopic BrdU incorporation in clones posterior to the second mitotic wave (Fig. 5A), while 0/17 mer mosaic eye discs tested had ectopic S-phases posterior to the second mitotic wave (Fig. 5B). We also detected ectopic S-phase entry in a 9/15 ex mosaic discs using a 30′ BrdU pulse, while all 19 mer discs tested using a 30′ BrdU pulse were negative. Similar results were obtained with independent alleles of ex (exe1 and exMGH2) and mer (mer4 and merF705), indicating that these findings are not allele-specific (data not shown). Thus, we conclude that ex clones have delayed cell cycle exit in the larval eye disc, while mer clones do not.
Consistent with the patterns of BrdU incorporation, mer and ex differentially regulate the patterns of Cyclin expression in the developing eye. We detected elevated Cyclin E levels in both ex clones and mer clones anterior to the morphogenetic furrow and in the second mitotic wave (Fig. 5C, D, horizontal yellow arrowheads). However, Cyclin E levels were strongly elevated posterior to the second mitotic wave in ex clones but only weakly or not at all in mer clones (Fig. 5C, D, horizontal white arrowhead). Ex also appears to regulate the levels of mitotic cyclins, as we observed elevated Cyclin B (Fig. 5E) and Cyclin A (data not shown) protein levels posterior to the morphogenetic furrow in ex clones in the eye disc. In contrast, we did not detect a change in Cyclin B levels in mer clones posterior to the morphogenetic furrow (Fig. 5F). Our data suggest that both ex and mer regulate Cyclin E levels in cells in the first and second mitotic waves. However, loss of ex results in ectopic mitotic cyclin levels and ectopic proliferation in the normally postmitotic cells posterior to the second mitotic wave, while, under identical conditions, loss of mer function does not. This suggests that loss of ex has a much greater role than mer in restricting Cyclin E expression and maintaining cells in a post-mitotic state in this portion of the eye disc.
Deregulation of the Hpo/Sav/Wts pathway results in increased expression of both Ex and DIAP1, (Hamaratoglu et al., 2006; Huang et al., 2005). Since both Ex and Mer are both thought to regulate the Hpo/Sav/Wts pathway, we examined how loss of either mer or ex affected Ex and DIAP1 protein levels. We found that Mer and Ex had differing effects on both DIAP1 and Ex protein levels. Loss of ex resulted in a robust upregulation of both Ex and DIAP1 protein levels across the entire eye imaginal disc (Fig. 6A, C). In contrast, loss of mer resulted in only a modest increase in Ex protein levels in the morphogenetic furrow (Fig. 6B). DIAP1 protein levels were upregulated only slightly, if at all, in mer mutant clones in the eye disc (Fig. 6D). Thus, there are significant differences in both the locations in which and/or the extent to which Mer and Ex regulate DIAP1 and Ex protein levels.
A subset of cells in the pupal retina is eliminated by apoptosis during retinal maturation (Cagan and Ready, 1989; Wolff and Ready, 1991). Antagonizing normal developmental apoptosis in the pupal retina either by loss of the pro-apoptotic gene hid or by expressing the baculovirus caspase inhibitor p35 (Clem et al., 1991) results in the appearance of supernumerary interommatidial cells in the mature pupal retina which are uniformly distributed and usually arranged in a single layer [Fig 7A’’’ and (Hay et al., 1994; Kurada and White, 1998)].
To determine whether mer or ex regulates normal developmental apoptosis in the pupal retina, we examined the cellular architecture of mature pupal retinas mosaic for either mer or ex. We observed a striking difference in the interommatidial cell phenotypes of ex and mer tissue in mature pupal retinas. ex tissue in mature pupal retinas appeared to have only a slight increase in the number of interommatidial cells (Fig. 7B) when compared to wild type tissue (Fig. 7A). In addition, ex mutant retinal tissue had an irregular ommatidial lattice; many ex ommatidia lacked a regular hexagonal shape and had extra bristle cells. In contrast to ex tissue, mer clones in mature pupal eye discs contained a large number of supernumerary interommatidial cells, but had a mostly regular hexagonal lattice. The supernumerary interommatidial cells in mer clones are largely arranged end-to-end and in a single layer (Fig. 7C).
To quantify the extent to which ex and mer tissue is defective in normal developmental apoptosis in the eye, we counted the number of cells present per ommatidial unit (see Materials and Methods for details of counting). ex clones in mature pupal retinas (Fig. 7B) contained slightly elevated numbers of cells per ommatidium when compared to control mosaic retinas (Fig. 7A); ex ommatidia (n=28) contained an average of 22.6 ± 1.7 cells per target area, while wild type ommatidia (n=26) contained an average of 21 ± 0.3 cells per target area. Though small, the difference between wild type and ex ommatidia is statistically significant (P<0.0001) and is consistent with a recent study showing that ex ommatidia contain approximately two more interommatidial cells per ommatidium (Silva et al., 2006). Strikingly, mer ommatidia contained 31.3 ± 3.0 cell per target area, a number significantly greater than either control or ex (P<0.0001 in both comparisons).
Because Wg signaling promotes a wave of early programmed cell death in the developing pupal retina (Cordero et al., 2004), we were curious whether ectopic Wg protein in ex mosaic eye discs promotes ectopic cell death that contributes to the irregular ommatidial lattice in ex mutant tissue (Fig. 7B) and might obscure an ectopic cell survival phenotype. When we examined mature pupal retinas doubly mutant for ex and wg, we found that reducing Wg function in ex tissue noticeably suppressed the irregular ommatidial patterning of ex mutant tissue but did not result in the appearance of more supernumerary interommatidial cells (Fig. 7D). This suggests that increased Wg signaling in some way contributes to irregular ommatidial architecture, but does not disguise a pronounced cell survival phenotype.
It is striking that relatively few interommatidial cells are present in ex tissue in mature pupal retinas, even though many additional interommatidial cells are generated in ex clones by delayed cell cycle exit during eye development. This implies that the apoptotic mechanism for elimination of supernumerary interommatidial cells is largely intact in ex tissue and is able to kill even more extra cells than are present in wild type tissue. In contrast, there are significant numbers of interommatidial cells present in mer tissue in the mature pupal retina, even though cell cycle exit in mer tissue in larval eye discs occurs in a timely manner. This suggests that Mer significantly regulates normal developmental apoptosis in the developing pupal eye, whereas Ex does not do so to any great extent.
Taken together, our data suggest that ex regulates cell cycle exit but regulates normal developmental apoptosis only slightly. In contrast, it appears that mer plays little or no role in cell cycle exit but regulates apoptosis in the pupal retina. To test this model, we antagonized apoptosis in pupal retinas mosaic for either mer or ex by expressing the caspase inhibitor p35 (Clem et al., 1991; Hay et al., 1994). Based on our initial model of ex and mer function, we predicted that inhibiting cell death in ex tissue would result in the appearance of a large number of supernumerary interommatidial cells, cells produced by delayed cell cycle exit (Fig. 5A) that are usually eliminated. In contrast, we predicted that inhibiting cell death in mer tissue would produce a phenotype similar to mer mutant tissue alone, as mer tissue does not appear to have a delay in cell cycle exit (Fig. 5B). Though antagonizing cell death in ex tissue produced large numbers of supernumerary interommatidial cells as predicted (Fig. 7B’’’), we were surprised to find that inhibiting cell death in mer tissue also resulted in the appearance of additional interommatidial cells (Fig 7C’’’). This suggests that mer tissue is only partially defective in normal developmental apoptosis, since the pupal retina has the capacity to eliminate a greater number of supernumerary cells than are produced in mer tissue, as occurs in ex (for example, see Fig. 7B). Consistent with this, we have been able to observe caspase-positive cells in mer tissue at the same time that the supernumerary interommatidial cells in the neighboring wild type tissue are dying (Fig. 7E). This suggests that loss of mer does not delay normal developmental apoptosis and that at least some mer interommatidial cells in the pupal retina are capable of dying.
Mer and Ex each regulate the intrinsic growth and proliferation rate of cycling cells. However, mer and ex clearly have distinct mutant phenotypes and functions. While Ex promotes appropriate cell cycle exit in the larval eye disc, Mer does not appear to do so. Conversely, while Mer promotes normal developmental apoptosis, Ex does not do so to any large extent. Finally, we have demonstrated that Ex has an additional function in regulating Wg protein levels in the eye that is not obviously shared by Mer or Hpo.
Loss of either mer or ex results in increased tissue growth. We have shown that there are significant qualitative and quantitative differences between the mer and ex single mutant phenotypes and that ex and mer each make distinct, non-redundant synergistic contributions to the ex mer double mutant phenotype. While both mer and ex negatively regulate growth in proliferating cells, they have distinct roles in regulating cell cycle exit and apoptosis.
Ex regulates cell cycle exit in the developing eye disc, but mer does not. We routinely detected ectopic S-phase entry in ex clones posterior to the second mitotic wave, indicating that cell cycle exit is delayed in ex mutant tissue. However, we did not observe a delay in cell cycle exit in mer clones in the eye imaginal disc. Consistent with this observation, ex clones posterior to the second mitotic wave have elevated levels of Cyclin E, Cyclin B, and Cyclin A. In contrast, mer clones posterior to the second mitotic wave have normal levels of Cyclin E and Cyclin B. Thus, ex appears to have a more important role than mer in regulating the exit from the cell cycle in the larval eye disc.
Although ex mutants produce extra cells posterior to the second mitotic wave, there are relatively few supernumerary interommatidial cells present in ex mosaic eyes when developmental apoptosis in the eye is complete. Additional cell proliferation posterior to the second mitotic wave results in an increase in the number of interommatidial cells, since spacing of the founding ommatidial cells, the R8 cells, has already occurred at the morphogenetic furrow (Ready et al., 1976; Tomlinson and Ready, 1987; Wolff and Ready, 1993). Thus, the mechanism for eliminating extra cells by apoptosis during the pupal stage must still be largely intact in ex tissue.
In contrast, mer clones in the mature pupal retina have significant numbers of supernumerary interommatidial cells. This implies that Mer function is involved in the elimination of extra interommatidial cells during eye development. However, mer mutant tissue is only partially defective in developmental apoptosis in the eye, since expressing the caspase inhibitor p35 in mer tissue results in the appearance of additional cells, cells that usually die in mer tissue. It is unclear why there are more extra interommatidial cells present in mer eye tissue expressing p35 than in wild type eye tissue expressing p35. One possibility is that mer tissue could experience a delay in cell cycle exit that is below the level of detection of our larval BrdU incorporation assays. Alternatively, mer cells could exit the cell cycle appropriately during the third larval instar, only to re-enter the cell cycle at some later point after the end of the larval stage of development. Finally, it is formally possible that neither loss of mer nor expression of p35 completely abolishes developmental apoptosis in the eye, and the combination of the two blocks death to a greater extent than either alone.
Thus, it is likely that the large numbers of extra cells observed in mer ex double mutant clones in the pupal retina are generated largely because they lack ex function and then fail to be eliminated because they lack mer function. Hence, if Mer and Ex function primarily through the Hpo/Sav/Wts pathway, then our data suggest that control of cell cycle exit and control of apoptosis are separable functions of the Hpo/Sav/Wts pathway.
Interestingly, the level of DIAP1 protein does not correlate well with the impairment of apoptosis during the pupal stage, since ex clones in the third instar eye disc have greatly elevated levels of DIAP1 protein, while the levels in mer clones are only slightly increased. This, combined with recent evidence that the Hpo/Sav/Wts pathway regulates the bantam micro-RNA, suggests that the impaired cell death in mer ex double mutant clones, as well as in clones mutant for hpo, sav or wts, may not be directly related to elevated DIAP1 protein levels and may involve another mechanism, such as regulation of the pro-apoptotic gene hid (Brennecke et al., 2003; Nolo et al., 2006; Thompson and Cohen, 2006; Udan et al., 2003).
Our experiments have uncovered a role for ex for regulating progression of the morphogenetic furrow. Ex inhibits Wg function, since ex mutant clones show elevated Wg protein levels and ectopic domains of Wg expression. As a result, eyes that are entirely composed of ex tissue are smaller than wild-type eyes, despite the overgrowth of the eye disc.
Although we have not observed ectopic Wg protein or delayed morphogenetic furrow progression in mer mutants, others have previously observed that reduction of mer function enhances the small eye phenotype of flies heterozygous for a strong allele of dpp (McCartney et al., 2000). The same group has also observed a photoreceptor differentiation defect similar to that seen with strong alleles of ex in eye tissue doubly mutant for a strong allele of mer and a weak allele of ex, but not for either mutation alone (Maitra et al., 2006). Both of these observations could be the result of a subtle, undetectable increase in Wg levels in mer tissue. Alternatively, Mer could regulate Wg signaling downstream of Wg protein levels.
Interestingly, we do not detect upregulation of Wg protein in hpo mosaic eye discs (Fig. 5). This suggests that regulation of Wg by ex in the eye could occur via a pathway other than the Hpo/Sav/Wts pathway. However, others have recently reported elevated Wg protein levels and an expansion of the domains of Wg protein in hpo clones, sav clones, wts clones, and mer ex double mutant clones in the wing imaginal disc (Cho et al., 2006). It is not yet clear why Hpo appears to regulate the Wg protein levels and domains in the wing but not in the eye. One possibility is that loss of hpo results in a much more subtle increase in Wg protein levels in the eye than in the wing. Alternatively, it is possible that Hpo regulates Wg in an organ-specific manner, while Ex regulates Wg protein more globally. Indeed, we have also detected elevated Wg protein levels and enlarged domains of Wg protein in ex clones in the third instar wing imaginal disc at the prospective wing margin, at the edges of the wing pouch, and in the hinge region (data not shown). Since Wg protein levels are regulated in a complex spatio-temporal pattern in eye and wing imaginal discs (Baonza and Freeman, 2002; Cavodeassi et al., 1999; Couso et al., 1993; Ng et al., 1996; Phillips and Whittle, 1993), widespread regulation of Wg protein levels by Ex suggests that Ex regulates Wg via a tissue and context-independent mechanism.
Our results suggest that the effects of increased Wg protein levels in the eye imaginal disc depend on the amount and distribution of Wg protein produced. Large increases in Wg protein in the eye, seen when a large ex clone occupies the dorsal or ventral margin of the eye disc, can significantly inhibit morphogenetic furrow progression in this portion of the eye disc and contribute to the small eye phenotype of ex homozygotes. When ex clones are more evenly spaced in the developing ex mosaic eye, ectopic Wg protein is more evenly distributed across the eye disc (data not shown). In this case, ectopic Wg seems to contribute to the eye overgrowth phenotype of ex mosaic eyes by modestly slowing morphogenetic furrow progression across the entire eye disc. This allows the tissue anterior to the morphogenetic furrow to proliferate for a longer period of time. Thus, increased Wg protein in the eye could contribute to overgrowth of the organ, even though changes in Wg levels do not affect the intrinsic ability of ex tissue to proliferate.
A prevailing model is that Ex and Mer function together and at least partly redundantly in some way upstream of the Hpo/Sav/Wts pathway. We have shown that Ex regulates Cyclin E expression and cell proliferation posterior to the second mitotic wave, whereas Mer regulates apoptosis in the pupal retina. The Hpo/Sav/Wts pathway can regulate both cell cycle exit and apoptosis, likely via the same transcriptional co-activator Yorkie. It is therefore puzzling that Expanded and Merlin could act through Yorkie to regulate the expression of distinct genes that regulate cell cycle exit in one case and apoptosis in the other. One possibility is that the modifications of Yorkie induced by Mer and Ex are somehow different such that they influence its preference for different promoters. An alternate possibility is that Mer and/or Ex can also regulate cell cycle exit and apoptosis independently of Yorkie. Biochemical studies that define the precise mode of interaction of the Hpo/Sav/Wts pathway and Yki with Mer and Ex will help address these issues.
We thank: W. Fowle for generating SEMs; T. Orr-Weaver, B. Hay, A. Laughon, J. Treisman, N. Dyson, R. Fehon, the Bloomington Stock Center, and the Developmental Studies Hybridoma Bank for antibodies and stocks. We also thank R. Krieser, L. Raftery, N.-S. Moon, K. Moberg, K. Harvey, A. J. O’Malley, and past and present members of the White and Hariharan labs for helpful discussions. B.P. was supported by the Dr. Jack A. Davis, M.D. postdoctoral fellowship from the American Cancer Society (PF-02-234-01) and by the Massachusetts Biomedical Research Council Tosteson postdoctoral fellowship. K.W: was supported in part by N.I.H. grant GM55568. I.K.H. was supported in part by N.I.H. grants GM61672 and CA95281
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.