dE2F2-Independent Rescue of Proliferation in Cells Lacking an Activator dE2F1

Abstract

In Drosophila melanogaster, the loss of activator de2f1 leads to a severe reduction in cell proliferation and repression of E2F targets. To date, the only known way to rescue the proliferation block in de2f1 mutants was through the inactivation of dE2F2. This suggests that dE2F2 provides a major contribution to the de2f1 mutant phenotype. Here, we report that in mosaic animals, in addition to de2f2, the loss of a DEAD box protein Belle (Bel) also rescues proliferation of de2f1 mutant cells. Surprisingly, the rescue occurs in a dE2F2-independent manner since the loss of Bel does not relieve dE2F2-mediated repression. In the eye disc, bel mutant cells fail to undergo a G₁ arrest in the morphogenetic furrow, delay photoreceptor recruitment and differentiation, and show a reduction of the transcription factor Ci155. The down-regulation of Ci155 is important since it is sufficient to partially rescue proliferation of de2f1 mutant cells. Thus, mutation of bel relieves the dE2F2-mediated cell cycle arrest in de2f1 mutant cells through a novel Ci155-dependent mechanism without functional inactivation of the dE2F2 repressor.

Cell proliferation and differentiation are precisely coordinated during the development of a multicellular organism. The loss of such control may ultimately lead to defects in development and cancer. E2F transcription factors are important regulators of the cell cycle and critical downstream targets of the retinoblastoma (pRB) tumor suppressor protein (12). In spite of remarkable progress, dissecting the pRB pathway in mammalian cells remains a challenging task. One issue arises from functional redundancy and compensation among E2F and pRB family members. In the past years, Drosophila melanogaster has been recognized as a valuable tool for understanding various aspects of E2F biology. This is mainly due to the high conservation in cell cycle regulation between flies and mammals and the relative simplicity of the E2F/pRB module in Drosophila. In Drosophila, there are two E2F genes. As mammalian counterparts, their products can be classified as repressors (dE2F2) and activators (dE2F1).

Analysis of dE2F single- and double-mutant animals has provided insights into the normal function of dE2Fs during development. de2f1 mutant larva are severely retarded in larval growth and show a strong reduction in S phases and the expression of E2F targets. Unexpectedly, these defects are largely suppressed in de2f1 de2f2 double mutants. This suggests that the de2f1 mutant phenotype is not entirely due to the absence of activator dE2F1 but rather, to some extent, is due to the presence of the “unchecked” repressor dE2F2 (18). This result is particularly striking since de2f2 mutants are viable and develop normally (5, 18). Why “unchecked” dE2F2 has such a strong effect on cell proliferation in the absence of dE2F1 is not clear. A possible explanation, supported by gene expression profiling (11), is that, in de2f1 mutant cells, dE2F2 inappropriately represses dE2F1-specific genes that are critical for cell proliferation. Until now, the only known way to rescue the de2f1 mutant phenotype in vivo was through mutations in de2f2 or rbf1 (13, 18, 41). This may indicate that de2f1 mutant cells are not capable of proliferating unless dE2F2/RBF is inactivated. However, this idea was not rigorously tested, and it is not clear whether cell cycle arrest in de2f1 mutant cells can be overridden in a dE2F2-independent manner.

These puzzling properties of the de2f1 mutant phenotype prompted us to take a genetic approach to address some of the unresolved questions about the mechanism of cell cycle arrest in these cells. Using a mosaic genetic screen in Drosophila, we found that the loss of DEAD box protein Belle (Bel) rescues the cell cycle defects of de2f1 mutant cells. Surprisingly, dE2F2-mediated repression is not relieved in bel mutants, suggesting that proliferation in de2f1 bel double-mutant cells occurs in the presence of a functional dE2F2/RBF. The results show that Bel specifically affects Hedgehog (Hh) signaling, and this effect is important for the rescue of the de2f1 mutant phenotype. Thus, cell cycle arrest in de2f1 mutant cells can be relieved not only in a dE2F2-dependent manner but also in a dE2F2-independent manner.

MATERIALS AND METHODS

Generation of the de2f2 FRT40A chromosome.

The de2f2 gene is located at 39B in close proximity to the centromere. Therefore, a de2f2 mutation cannot be combined with FRT40A by standard meiotic recombination. Additionally, since de2f2 mutant alleles are viable, new de2f2 mutations cannot be easily induced on the FRT40A-containing chromosome. To overcome these problems, we employed P element-induced male recombination. With low frequency, a P element transposase generates crossovers in the P element in the male germ line. Since FRT40A is a P element insertion, we used it to recover rare crossover events on FRT40A. The de2f2^c03344 mutant allele is a PB insertion into the de2f2 gene. The advantage in using this allele is that PB is stable in the presence of the P element transposase. Briefly, a dp de2f2^c03344 bw chromosome that carries recessive markers to identify the recombinants was constructed. Potential recombination events were revealed in the progeny of w; dp de2f2^c03344 bw/dp⁺ FRT40A bw⁺; Δ2-3 Sb/TM6 Tb males by segregation among dp and bw markers. Two independent stocks of w; dp de2f2^c03344 FRT40A flies were recovered.

Fly stocks.

The following mutant alleles were used: de2f1 mutant, de2f1⁷²⁹ (16); smo mutant, strong hypomorphic mutation smo^IIG26 (46); bel mutant, bel^L4740; de2f2 mutants, de2f2^c03344; dDP mutant, dDP^a3.

bel^7d19 was isolated in the FLP/FLP recombination target (FRT) screen for suppressors of the de2f1 mutant phenotype. FRT82B de2f1⁷²⁹ males were mutagenized with ethyl methanesulfonate and crossed to w; TM3/MKRS females, and then male progeny were crossed individually to y w ey-FLP; FRT82B P[mini-w]90E l(3)cl-R3¹/TM6B Tb females. Since clones of de2f1 mutant cells are very small, the majority of flies had no visible white spots marking de2f1 mutant cells. Exceptional flies, whose eyes contained large white patches indicating that the block in de2f1⁷²⁹ mutant cells is overridden, were retained. Forty thousand chromosomes were screened in the F2 screen. Mapping was performed by crossing 7d19 flies to flies with a set of deficiencies uncovering 3R. 7d19 fails to complement deficiency Df(3)Exel6149 and several alleles of the gene bel; the gene bel was uncovered by the deficiency.

For the eye disc analysis, clones of homozygous mutant cells were generated in the animals of the following genotypes: ey-FLP; FRT82B bel^7d19 de2f1⁷²⁹/FRT82B P[Ubi-GFP], ey-FLP; FRT82B bel^L4740 de2f1⁷²⁹/FRT82B P[Ubi-GFP], ey-FLP; FRT82B bel^L4740/FRT82B P[Ubi-GFP], ey-FLP; FRT40A de2f2^c03344/FRT40A P[Ubi-GFP], ey-FLP; FRT40A de2f2^c03344/FRT40A P[Ubi-GFP]; FRT82B de2f1⁷²⁹/FRT82B P[Ubi-GFP], ey-FLP; FRT40A smo^IIG26/FRT40A P[Ubi-GFP]; FRT82B de2f1⁷²⁹/FRT82B P[Ubi-GFP], ey-FLP; 2x PCNA-GFP; FRT82B bel^7d19 de2f1⁷²⁹/FRT82B P[Ubi-GFP], and ey-FLP; FRT42D dDP^a3/FRT42D P[Ubi-GFP].

Immunofluorescence.

Eye imaginal discs were dissected in 10% fetal bovine serum-Schneider's insect medium and then fixed in 4% formaldehyde-phosphate-buffered saline (PBS). Washes were then done using 0.3% Triton X-100-PBS. Discs were then blocked in 0.1% Triton X-100-10% normal donkey serum-PBS for 1 h before addition of appropriate primary antibodies. Washes were then done using 0.1% Triton X-100-PBS. Discs were then incubated in blocking solution containing the appropriate secondary antibodies. The secondary conjugated antibodies used in this work were anti-mouse Alexa Fluor 405 (1:500; Invitrogen), anti-rabbit Alexa Fluor 405 (1:500; Invitrogen), anti-mouse Cy3 (1:100), anti-rabbit Cy3 (1:100), anti-rabbit fluorescein isothiocyanate (1:200), and anti-rat Cy3 (1:50) from Jackson Immunolaboratories. Washes were then done using 0.1% Triton X-100-PBS. Bromodeoxyuridine (BrdU) labeling of eye imaginal discs was performed as previously described (17, 41). Images were collected on a Zeiss LSM510 confocal microscope. The primary antibodies used in this work were rabbit anti-β-galactosidase (β-Gal; 1:500; Cappel), mouse anti-BrdU (1:50) (BD Bioscience), rabbit anti-green fluorescent protein (GFP; 1:1,000) (Invitrogen), rabbit anti-phospho-H3 (1:150) (Upstate), mouse anti-RBF1 (14), mouse anti-cyclin E (1:5) (42), rabbit anti-dE2F2 (1:100) (18), rabbit anti-phosphorylated Mothers against decapentaplegic (pMad; 1:5,000) (39), rat anti-Ci155 (1:1) (36), rabbit anti-Atonal (Ato; 1:50) (25), and rabbit anti-Bel (1:500) (26). Mouse antibodies anti-cyclin B (1:10), anti-Notch (1:100), anti-Delta (Dl; 1:50), anti-Pros (1:20), anti-Patched (Ptc; 1:50), anti-Cut (1:10), and anti-β-Gal (1:20) and rat anti-Elav (1:50) are from the Developmental Studies Hybridoma Bank.

Real-time PCR.

Quantitative real-time PCR on RNA isolated from S2 cells was performed as previously described (41). To measure the expression of Arp53D in the eye imaginal discs, total RNA was isolated from wild-type and mutant discs of different genotypes. Since Arp53D is not expressed in wild-type cells, we expect that the presence of RNA from these cells in the sample would not obscure derepression of Arp53D in the mutant cells. This was confirmed by using RNA prepared from the eye discs containing clones of dDP mutant cells. dDP represents a valid control since Arp53D is fully derepressed in dDP-deficient cells (11) and clones of dDP mutant cells are small (19). RNA was prepared from the eye imaginal discs dissected from wild-type larvae, dDP mutant larvae (in which essentially all cells are dDP mutant), and larvae in which clones of dDP mutant cells were induced using ey-FLP (where approximately 10% of all cells are mutant and 90% are wild type [19]). The results of real-time PCR are shown (see Fig. 4B). The Arp53D signal is increased by a factor of several thousand in RNA isolated from the eye imaginal discs of dDP homozygous mutant larvae and is increased by a factor of several hundred in an RNA sample isolated from the mosaic eye imaginal discs carrying clones of dDP mutant cells. Thus, this technique is sensitive and reliable to detect the expression of Arp53D is a small subset of cells.

FIG. 4.

Effect of Bel on dE2F2-mediated repression. (A) The loss of bel fails to restore the normal level of PCNA-GFP reporter (green) in the SMW inhibited in de2f1 mutant cells. bel de2f1 double-mutant cells are marked by the presence of β-Gal (red), and wild-type cells are distinguished by the lack of β-Gal. The position of the MF is shown by the arrowhead. (B) A well-known dE2F2-specific target, Arp53D, remains fully repressed in the eye discs containing clones of bel single-mutant and bel de2f1 double-mutant cells. Arp53D is strongly derepressed in dDP mutant eye discs (DP) and in the eye discs containing clones of dDP mutant cells (FLP DP). The level of Arp53D was determined by real-time reverse transcription-PCR. The means of the three experiments are shown above the bars. (C) Depletion of Bel in tissue culture cells has no effect on the ability of dE2F2 to repress a PCNA-luc reporter in transient transfections. Cells were incubated with control (solid bars) or Bel (open bars) double-stranded RNA (dsRNA). After 4 days, cells were transfected with a PCNA-luc reporter and with either an empty vector or pIE4-dE2F2 expression plasmids in triplicate. To normalize for the efficiency of transfection, a β-Gal expression plasmid was included. Depletion of Bel was verified by Western blotting. (D) Bel does not repress the E2F reporter in transient transfections. Cells were cotransfected with a PCNA-luc reporter; a β-Gal expression plasmid; and vector alone or a pIE-dE2F2, pIE4-dE2F1, or pIE-S-Bel expression construct. As expected, dE2F2 represses the E2F reporter while dE2F1 activates it. In contrast, overexpression of Bel does not repress the PCNA-luc reporter. Expression of exogenous Bel was verified by Western blotting as shown below.

Manipulations with S2 cells.

RNA interference (RNAi), transfections, and Western blotting were done as previously described (41). The primary antibodies used in this work were rabbit anti-Bel (1:6,000) (26), rabbit anti-dE2F2 (1:2,000) (18), and mouse anti-RBF1 (DX5; 1:5) (14).

Using the pIEx-7 Ek/LIC vector (Novagen), we cloned in the wild-type Bel cDNA open reading frame (ORF; Drosophila Genomics Resource Center). The construct features a C-terminal S tag for protein detection. This expression construct was sequenced to confirm accurate insertion of the Bel ORF.

RESULTS

The loss of de2f2 rescues the proliferative defect of de2f1 mutant cells in mosaic animals.

As has been previously shown, clones of de2f1 mutant cells are extremely small (3, 37, 41). Although the larval phenotype of the de2f1 mutant is largely rescued by mutation of de2f2, it has not been tested whether and to what extent the loss of de2f2 suppresses the small clone size of de2f1 mutant cells. To date, a de2f2 FRT mutant chromosome has not been reported. We used a novel strategy to create a de2f2 FRT40A mutant chromosome that carries loss-of-function allele de2f2^c03344. In the eye discs, clones of de2f2 mutant cells can be distinguished by the lack of a GFP marker while the wild-type tissue expresses GFP. As expected, the dE2F2 protein is undetectable in clones of de2f2^c03344 mutant cells (Fig. 1B and B *).

FIG. 1.

Rescue of proliferation in clones of de2f1 mutant cells by mutation of de2f2. Eye imaginal discs were dissected from third-instar larva. Clones of mutant cells were induced with ey-FLP and are identified by the lack of GFP (green). Posterior is to the right. Genotypes are shown. (A) Control experiment showing clones of homozygous tissue with the wild-type chromosome. In the absence of mutations, there are approximately equal amounts of tissue marked by GFP and by the lack of GFP. (B) Clones of de2f2^c03344 mutant cells (GFP negative) do not express dE2F2 protein. (C, C*, D, and D*) Induction of clones of cells that are homozygously mutant for de2f2^c03344 and de2f1⁷²⁹. Double-mutant cells fail to express GFP. Cells that are homozygously mutant for the de2f1⁷²⁹ allele express β-Gal (red) from the PZ[lacZ] element inserted into the de2f1 gene. Note that large clones of de2f1 homozygous mutant cells (red) can be recovered only in de2f2 homozygous mutant cells (lack of GFP), while, in the presence of de2f2 (GFP positive), clones of de2f1 mutant cells do not exceed one to three cells (yellow). (C*) GFP clone marker alone of the same image as in panel C. (D and D*) Higher magnification of the same image as in panels C and C*. Two clones of de2f1 de2f2 double-mutant cells are outlined. Arrows point to de2f1 single-mutant cells that carry wild-type de2f2. Staining with DNA dye 4′,6′-diamidino-2-phenylindole confirms that these are individual cells. (D*) GFP clone marker of the same image as in panel D.

Next, we induced clones of de2f1 de2f2 double-mutant cells. Since de2f1 and de2f2 are on different chromosomes, partially overlapping clones of de2f1 single-, de2f2 single-, and de2f1 de2f2 double-mutant cells are recovered. Cells that are de2f1 homozygous mutants are marked by β-Gal, while de2f2 de2f1 double mutants can be distinguished by the lack of GFP signal. Strikingly, relatively large clones of de2f1 mutant cells (β-Gal positive) can be recovered only when de2f2 is inactivated (GFP negative) (Fig. 1C, C*, D, and D*). In contrast, clones of de2f1 mutant cells (β-Gal positive) are extremely small in the presence of the wild-type copy of de2f2 (GFP positive). Thus, proliferation of de2f1 mutant cells is rescued by the loss of de2f2 in mosaic animals. We note that the rescue is not complete and that the clones of de2f1 de2f2 double-mutant cells are smaller than clones of wild-type cells (compare Fig. 1C and A). Thus, E2F-deficient cells are at a proliferative disadvantage when directly compared with wild-type cells. This observation supports the idea that E2F-deficient cells are not completely normal (19, 35) and further underscores the importance of E2F regulation for normal proliferation.

The loss of bel rescues proliferation of the de2f1 mutant cells.

We have used the FLP/FRT system to systematically screen the Drosophila eye for genes that when mutated rescue proliferation of de2f1 mutant cells. One mutant isolated in the screen was named 7d19. The small size of clones of de2f1 mutant cells (3, 37) was rescued by 7d19, and numerous clones of 7d19 de2f1⁷²⁹ double-mutant cells were recovered (Fig. 2A). By standard mapping techniques, 7d19 has been mapped to the bel gene. bel encodes a DEAD box RNA helicase that is highly conserved from yeast to humans. DEAD box proteins have been implicated in RNA-processing events such as splicing, nuclear export, degradation, and translation (9, 29, 31). Sequencing of the bel ORF from the 7d19 chromosome identifies a point mutation that results in a glycine-to-serine substitution at position 414. This mutation affects the highly conserved “GG doublet” of the DEAD box protein family (9). Mutation of either of these glycines completely abolishes the function of the related DEAD box helicase eukaryotic initiation factor 4A (eIF4A) in yeast (9). Thus, the bel^7d19 mutant appears to be a strong loss-of-function allele and accordingly it fails to complement known bel mutant alleles. To confirm the identity of bel, we tested whether a known loss-of-function allele of bel, bel^L4740, rescues the de2f1 mutant phenotype. In the bel^L4740 mutant allele, a P element is inserted into the bel gene and no protein is produced (Fig. 2B and B*). Numerous clones of bel^L4740 de2f1⁷²⁹ double-mutant cells were recovered (Fig. 2C). Thus, the genetic interaction between de2f1 and bel mutant alleles strongly suggests that the loss of bel relieves the cell cycle arrest of de2f1 mutant cells.

FIG. 2.

The loss of bel rescues proliferation of de2f1 mutant cells. Wild-type cells are marked by the presence of GFP (green), while mutant tissue lacks GFP. Clones of mutant cells were induced with ey-FLP and are identified by the lack of GFP (green). (A) Clones of bel^7d19 de2f1⁷²⁹ double-mutant cells. (B and B*) The loss-of-function bel^L4740 mutant allele makes no Bel protein (red), as revealed by the absence of staining with anti-Bel antibody in clones of bel^L4740 mutant cells (absence of green). (B*) The same image as in panel B, showing expression of Bel (red) without GFP. (C) The known loss-of-function bel^L4740 mutant allele rescues proliferation of de2f1 mutant cells. Note the appearance of patches of double-mutant, GFP-negative tissue. (D) The eye discs were labeled with BrdU (red) to visualize S phases. The MF is marked by the arrowhead. Wild-type cells asynchronously cycle anterior to the MF, are arrested in G₁ within the MF, and synchronously enter S phase in the SMW posterior to the MF. The loss of bel partially rescues the SMW in de2f1 mutant cells, as evident by the appearance of bel de2f1 double-mutant BrdU-positive cells in the SMW. (E) Distribution of Bel protein in a wild-type eye disc. Bel is shown in red, and DAPI staining is shown in blue. Bel exhibits a predominantly cytoplasmic distribution.

In the eye disc, the patterns of S phases can be visualized by BrdU labeling. During larval development, a morphogenetic furrow (MF) sweeps from posterior to anterior across the eye imaginal disc. As asynchronously dividing cells enter the MF, they undergo a G₁ arrest. Posterior to the MF uncommitted cells enter a final synchronous cell division called the second mitotic wave (SMW), visualized as a sharp band of BrdU-positive cells. The loss of de2f1 prevents entry into S phase in the SMW (13). Mutation of bel partially rescues this defect. As shown in Fig. 2D, bel de2f1 double-mutant cells enter S phase in the SMW although the number of BrdU-positive cells and the overall level of BrdU labeling are somewhat reduced compared to those for the wild type.

As an initial step in characterizing Bel, we immunostained eye discs with anti-Bel antibody. In the wild-type tissue, Bel protein is expressed ubiquitously throughout the disc. This staining is highly specific because it is absent in cells homozygous for the bel^L4740 mutant allele, which is expected to make no Bel protein (Fig. 2B and B*). High-magnification confocal images show that Bel is absent from the nucleus and localized to the cytoplasm, as would be expected for an RNA helicase acting posttranscriptionally (Fig. 2E). We repeatedly observed a slight but reproducible increase of Bel protein in cells preceding the MF, while its expression is somewhat reduced in the more posterior differentiating cells.

The effect of belle on dE2F2-mediated repression.

One of the hallmarks of the de2f1 mutant phenotype is a reduction of E2F-dependent transcription, which is partially due to dE2F2-mediated repression. Therefore, we asked whether bel affects dE2F2-mediated repression. The trivial explanation that the levels of RBF1 or dE2F2 are down-regulated by mutation of bel is unlikely since both proteins are expressed normally in bel mutant cells and in RNAi-treated tissue culture cells (Fig. 3). In the eye discs, a PCNA-GFP reporter construct is commonly used as a sensitive readout of E2F transcription (45). This reporter reflects the normal expression of PCNA, a well-characterized E2F target, and has been previously validated to assess E2F transcriptional activity (2, 45). We examined the expression of the reporter in the eye imaginal discs following induction of clones of bel de2f1 double-mutant cells. The reporter is expressed at a high level ahead of the MF and in a stripe of cells immediately posterior in wild-type cells. However, the PCNA-GFP reporter remains repressed in bel de2f1 double-mutant cells (Fig. 4A). Thus, although mutation of bel rescues the proliferation of de2f1 mutant cells, it appears that the loss of bel fails to restore the expression of the E2F reporter in de2f1 mutant cells. This raises the possibility that dE2F2-mediated repression is not affected in bel mutant cells. This was confirmed in two ways. First, we measured the expression of a known dE2F2-specific target, Arp53D, in the eye imaginal discs by real-time reverse transcription-PCR (for details see Materials and Methods). In wild-type cells, Arp53D is expressed at very low levels but is strongly derepressed in cells lacking dE2F2 or dDP both in vivo and in vitro (11). In bel mutant cells, the level of Arp53D transcripts is as low as in wild-type cells, indicating that Arp53D remains fully repressed (Fig. 4B). Similarly, depletion of Bel by RNAi in Drosophila S2 tissue culture cells does not relieve the repression of Arp53D (data not shown). Second, we examined the effect of Bel on dE2F2-mediated repression in transient transfections. In agreement with the previous results, dE2F2 represses an E2F reporter equally well in both control and Bel-depleted cells (Fig. 4C). Accordingly, the overexpression of Bel in transient transfections fails to repress the E2F reporter (Fig. 4D).

FIG. 3.

Levels of dE2F2 and RBF1 are normal in bel mutant cells. Clones of mutant cells were induced with ey-FLP and are identified by the lack of GFP (green). (A, A*, B, and B*) Clones of bel mutant cells were induced in the eye discs, and the levels of RBF1 and dE2F2 were determined by immunofluorescence. Clones of mutant cells are distinguished by the lack of GFP. (A and A*) RBF1 (red) is expressed ubiquitously in the eye disc, with a slight increase anterior to the MF. Staining with anti-phospho-H3 antibody (blue in panel A) marks the positions of the MF (arrowhead) and the SMW. (A*) The same image as in panel A, showing phospho-H3 (white) and RBF1 (red) expression. (B and B*) dE2F2 (red) is expressed normally in clones of bel mutant cells (lack of GFP). (B*) The same image as in panel B, showing dE2F2 expression (red). (C) Bel was depleted by RNAi in S2 cells, and the levels of RBF1, cyclin A (CycA), CycB, and dE2F2 were determined by Western blot analysis. A nonspecific band for dE2F2 is marked by an arrow. Tubulin was used as a loading control.

Taken together, these data indicate that mutation of bel does not generally relieve E2F-mediated repression. Although we cannot exclude the possibility that some E2F targets are derepressed in bel mutants, the explanation that the loss of bel rescues the de2f1 mutant phenotype through a global release of E2F-dependent repression is unlikely. The inability of bel mutations to restore the transcriptional defects in de2f1 mutant cells may explain why the loss of bel provides only a partial rescue of the de2f1 mutant phenotype.

bel is required for G₁ arrest in the MF in the developing eye.

The results described above suggest that bel rescues the de2f1 mutant phenotype in a dE2F2-independent manner. To gain insights into the mechanism of the rescue, we examined bel mutant cells in greater detail. In particular, we focused on the region of the eye disc known as the MF, where wild-type cells are arrested in G₁ prior to entering the SMW (Fig. 5A and B). Strikingly, while neighboring wild-type cells synchronously undergo G₁ arrest, bel mutant cells continue cycling. This is evident by the presence of bel mutant cells, which ectopically incorporate BrdU (Fig. 5C), show an abnormally high level of mitotic cyclin B (Fig. 5D) and cyclin A (data not shown), and undergo mitoses in the anterior part of the MF (Fig. 5E and E*). Double staining with cyclin B and phospho-H3 revealed the presence of cells that have a high level of cyclin B but were phospho-H3 negative, cells that a have high level of cyclin B and phospho-H3, and cells that had already degraded cyclin B but retained a high level of phospho-H3 (Fig. 5E and E*). The random distribution of the cell cycle markers suggests that bel mutant cells continue to asynchronously cycle in the MF. However, bel mutant cells do eventually arrest in the posterior part of the MF. These results indicate that bel promotes a G₁ arrest, at least in this developmental context. In an attempt to test this in overexpression experiments, we used an EP element located immediately upstream of the bel gene to drive ectopic Bel expression with different Gal4 drivers. However, the level of exogenous Bel protein was low and insufficient to affect the cell cycle progression, presumably due to a high level of the endogenous Bel protein.

FIG. 5.

bel mutant cells fail to arrest in G₁ in the MF. Clones of mutant cells are identified by the lack of GFP (green). (A and B) In wild-type discs, cells are arrested in G₁ in the MF (arrowhead). This is evident by the lack of cells that are BrdU positive (red) and express mitotic marker phospho-H3 (blue) (A) and the absence of cells that express the G₂ marker cyclin B (CycB; red) (B) within the MF. The image shown in panel A is at a higher magnification than the image in panel B. Anterior to the MF, the number of mitotic cells is increased. In the SMW, a synchronous wave of S phases is followed by a wave of mitoses. (C, D, E, and E*) bel mutant cells (GFP negative) fail to arrest in the anterior region of the MF and continue cycling, as evident by the appearance of mutant cells that show ectopic markers of S phase (BrdU), G₂ (CycB), and mitosis (phospho-H3). (C) bel mutant cells ectopically incorporate BrdU (red) in the MF. (D) bel mutant cells ectopically express CycB (red; arrow) in the MF. (E) bel mutant cells show ectopic CycB (red) and phospho-H3 (blue) in the MF. Double staining with CycB and phospho-H3 reveals cells that express only CycB (red; arrowhead), CycB, and phospho-H3 (magenta; thick arrow) and cells that degraded CycB but still express phospho-H3 (blue; thin arrow). Random distribution of the cell cycle markers indicates that bel mutant cells are asynchronously cycling. (E*) The same image as in panel E, showing the GFP clone marker (green) and CycB (red). The position of a clone is outlined. (F and G) bel mutant cells arrest normally following gamma irradiation. Clones of cells homozygous for wild-type (F) and bel^L4740 mutant (G) chromosomes were induced with ey-FLP. Larvae were irradiated and kept at 25°C for 1 h, and mitotic cells were visualized with anti-phospho-H3 antibody.

We sought additional evidence that bel mutant cells asynchronously proliferate in the MF, instead of being arrested randomly in the cell cycle. In the wild-type eye imaginal disc, proliferating cells respond to gamma irradiation by arresting in G₂/M in response to a DNA damage checkpoint. This arrest can be visualized by complete disappearance of a mitotic phospho-H3 marker 1 h after irradiation (Fig. 5F). We reasoned that, if bel mutant cells are cycling in the MF, then there will be no phospho-H3-positive cells following gamma irradiation. Conversely, if, in the MF, bel mutant cells are arrested randomly within the cell cycle, then we expect to detect the persistence of cells expressing phospho-H3 after gamma irradiation. After 1 h of irradiation, phospho-H3-positive cells were completely absent among mutant and wild-type cells (Fig. 5G and F). From this result we come to two conclusions. First, a DNA damage checkpoint remains intact in bel mutants. Second, bel mutant cells fail to arrest in G₁ in the MF and indeed asynchronously cycle.

The Cubitus interruptus transcription factor, Ci155, is down-regulated in bel mutant cells.

To gain insight into the normal function of bel in wild-type cells, we focused on characterization of the failure of bel mutant cells to arrest within the MF. The G₁ arrest in the MF has been shown to be dependent on the partially redundant Hh and decapentaplegic (Dpp) pathways (2, 17, 23, 47). Therefore, we examined whether these pathways are deregulated in bel mutant cells. First, we determined the expression of pMad, an activated form of a downstream transcription factor of Dpp signaling in the eye (1, 40). In wild-type cells, pMad is expressed at a low level near the MF, with a characteristic granular distribution and increases in regular groups of posterior cells (Fig. 6A and A *) (47). In clones of bel mutant cells, both high- and low-level expression of pMad antigen is normal.

FIG. 6.

Mutation of bel specifically reduces the level of Ci155. Clones of bel mutant cells were induced with ey-FLP are marked by the absence of GFP (green). Panels marked with an asterisk lack the GFP signal, and clones of bel mutant cells are outlined. (A and A*) A marker of the Dpp pathway, pMad (red in panel A and white in panel A*), is expressed in bel mutant cells. (B and B*) Expression of Ci155 (red in panel A and white in panel A*) is strongly reduced in bel mutant cells. (C and C*) Ptc is expressed in differentiating ommatidia and at a low level in the MF (marked with a bar under the image). Ptc (red in panel C and white in panel C*) is strongly down-regulated in the MF in bel mutant cells. (D and D*) Expression of Notch (N) (red in panel D and white in panel D*) is not altered in bel mutant cells. (E and E*) Expression of the Notch ligand, Dl (red in panel E and white in panel E*), is normal in bel mutant cells. (F and F*) Expression of p-ERK, a marker of activated EGFR (red in panel F and white in panel F*), is unaffected by a mutation in bel. (G) Mutation of smo partially rescues the small clone size of de2f1 mutant cells. smo de2f1 double-mutant cells are distinguished by the lack of GFP (green). (H and I) The effect of Bel on transcriptional activation of a Ptc reporter gene in Drosophila S2 cells. (H) Ci155 fails to efficiently activate the reporter in Bel-deficient cells. Control cells (solid bars) or cells treated with Bel double-stranded RNA (dsRNA) (open bars) were transfected with an empty vector or pDA-Ci plasmid (Ci) together with a luciferase reporter plasmid under the control of either a wild-type (ptcΔ136-Luc) or mutant (ptcΔ136-mut) Ptc promoter. To determine transfection efficiency, a β-Gal expression plasmid was cotransfected in each well. The experiments were done in triplicate to calculate the averages and standard errors. (I) Bel potentiates Ci155-dependent activation of the Ptc reporter construct. Transfections were done essentially as in panel H but without an RNAi treatment, and an S-Tag-Bel expression plasmid was included. Expression of S-Tag-Bel was confirmed by Western blotting.

The full-length form of transcription factor Ci155 is a downstream nuclear effector of the Hh pathway and commonly used as a readout of activation of Hh signaling (33, 36, 47). In wild-type cells, Ci155 accumulates to a high level anterior to the MF. However, the level of Ci155 is strongly reduced in clones of bel mutant cells (Fig. 6B and B*). Ptc is a well known target of Ci155. In the eye imaginal disc, Ptc is expressed at a low level in a thin stripe in the MF and at a higher level in differentiating photoreceptors (43). In agreement with the reduced level of Ci155, Ptc is strongly down-regulated in the MF in clones of bel mutant cells (Fig. 6C and C*). The reduction of Ci155 is not simply a consequence of ectopic proliferation of bel mutant cells in the MF since Ci155 has been previously shown to be expressed normally in thick veins mutant cells that also fail to arrest in the MF (17, 47).

We next tested whether the loss of bel specifically reduces Ci155 or whether it has a more general effect on other developmental pathways. We focused on Notch and epidermal growth factor receptor (EGFR) signaling due to their critical role in cell proliferation, differentiation, and survival during eye development. In wild-type cells, Notch is expressed in the MF and the SMW. As shown in Fig. 6D, D*, E, and E*, the pattern of Notch expression and that of its ligand Dl is unaffected in bel mutant cells. To assess activation of the EGFR pathway, we looked at the distribution of phospho-extracellular signal-regulated kinase (p-ERK). p-ERK is expressed in the MF in a group of cells from which the R8 photoreceptor cell is selected. p-ERK staining was observed in clones of bel mutant cells, indicating that the loss of bel does not affect the EGFR pathway (Fig. 6F and F*). Thus, the loss of bel specifically reduces Ci155, the effector of Hh signaling, but has no effect on several other tested pathways.

Taken together these data raise a question as to whether the reduced level of Ci155 in bel mutant cells is important for the rescue of cell proliferation in de2f1 mutant cells. To test this idea, we introduced a smoothened (smo) mutation into de2f1 mutants since smo mutant cells fail to accumulate Ci155. The mutation of smo rescues the proliferation of de2f1 mutant cells, as evident by the appearance of multiple clones of smo de2f1 double-mutant cells, although the rescue is slightly less efficient than in the case of bel mutations (Fig. 6G). Thus, we concluded that the reduction of Ci155 by bel mutations is an important mechanism in relieving the cell cycle block in de2f1 mutant cells.

Next, we examined the effect of Bel on Ci155-mediated transcriptional activation in transient transfections in S2 cells. We used a ptcΔ136-Luc reporter, which contains a patched promoter fused to luciferase. Although Ci155 is not expressed endogenously in S2 cells, the cells retain a normal response to Hh signaling once Ci155 is provided exogenously (20, 34). In control-treated cells, transfection of the Ci155 expression construct activates the reporter (Fig. 6H). This activation is dependent on a Ci155 binding site because a mutant reporter construct, ptcΔ136-mut, does not respond to Ci155. In Bel-depleted cells, Ci155 activation is significantly reduced. This is not due to an effect of Bel on the basal level of the reporter since the expression of a mutant reporter is unaffected in Bel-depleted cells. These results are consistent with the reduced level of Ptc in clones of bel mutant cells (Fig. 6C). In a converse experiment, overexpression of Bel cooperates with Ci155 and increases the expression of the reporter (Fig. 6I). It has been reported that the known components of the Hh pathway modulate the expression of the ptcΔ136-Luc reporter to a similar magnitude in transient transfections (20). Thus, our data indicate the importance of Bel in Hh signaling.

Differentiation is delayed in bel mutant cells in the eye imaginal disc.

Next, we examined photoreceptor differentiation in bel mutant cells. Ato is a transcription factor that determines the first photoreceptor, R8, to differentiate (25). In bel mutant cells, both up-regulation and resolution of Ato expression are delayed (Fig. 7A and A *). We used markers for different cell types to further define the differentiation defects in bel mutant cells. We find that expression of a neuronal marker, Elav, is delayed and displaced slightly posteriorly in bel mutant cells (Fig. 7B and B*). This is in agreement with the delay in R8 recruitment, as revealed by Ato expression. In more posteriorly situated clones, most of the bel mutant cells eventually express Elav, although there are some cells that remain Elav negative. Staining for cone cell markers Prospero and Cut revealed that there are very few cone cells present in clones of bel mutant cells and that cone cell differentiation is severely delayed (Fig. 7C, C*, D, and D*). Thus, we concluded that bel mutant cells delay, but still undergo, differentiation and that this delay is likely to be a consequence of delayed recruitment of ommatidial cells.

FIG. 7.

The loss of bel delays the onset of differentiation. Clones of bel mutant cells were induced with ey-FLP and are distinguished by the lack of GFP (green). Cells are stained with an R8 marker anti-Ato (A), neuronal marker anti-Elav (B), and the cone cell markers anti-Prospero (C), and anti-Cut (D), shown in red. Panels marked with an asterisk lack the GFP signal.

DISCUSSION

In an unbiased mosaic genetic screen, we have identified bel as a suppressor of the de2f1 mutant phenotype. Although several “cell cycle” screens in Drosophila have been conducted, the bel gene was not among the isolated modifiers (4, 27, 44, 48). This may be due to the limitations of the previous screening strategies, which relied mostly upon overexpression of dE2F or RBF1 to generate the phenotype. In this work, we combined the FLP/FRT technique with the de2f1 loss-of-function phenotype. This allowed us to rapidly screen for randomly induced recessive suppressor mutations while avoiding the unwanted effects provoked by the proteins present at abnormally high levels.

Previously, the de2f1 larval phenotype was shown to be rescued by mutating the de2f2 gene. Similarly, in mosaic animals, the cell cycle block in de2f1 mutant cells is also dependent on de2f2, and the clones of cells lacking dE2Fs can be recovered even when competing with wild-type cells. Surprisingly, bel rescues the de2f1 mutant phenotype through a novel, dE2F2-independent mechanism. In bel mutant cells, the levels of dE2F2 and RBF1 are normal and bel mutations do not generally relieve dE2F2-mediated repression. To our knowledge, this is the first example of proliferation of de2f1 mutant cells occurring in vivo in the presence of a functional dE2F2/RBF1 complex. In spite of the inability to relieve dE2F2-mediated repression, the loss of bel rescues the cell proliferation defects of de2f1 mutant cells. In the eye discs, bel mutations partially restore S phase entry in the SMW, which is inhibited in de2f1 mutant cells. However, mutant cells proliferate slower than their wild-type neighbors since the clones of de2f1 bel double-mutant cells are relatively small. This is consistent with the idea that dE2F-deficient cells are defective in their proliferation potential (19; this work). Alternatively, the loss of bel might not fully rescue all of the defects in de2f1 mutant cells, including those that are dE2F2 dependent, since bel mutants fail to relieve the dE2F2-mediated repression. Nevertheless, de2f1 bel double-mutant cells progress through the cell cycle and differentiate, indicating that the mutant cells are largely normal in their response to developmental cues.

How does bel rescue the cell cycle defects in de2f1 mutant cells? bel mutant cells fail to arrest in G₁ in the anterior part of the MF. Previous work has clearly established the central role of Dpp and Hh signaling in promoting cell cycle arrest in the MF (2, 17, 23, 47). In bel mutant cells, the expression of the downstream effector of Hh signaling Ci155 is strongly reduced. Consistently, mutation of smo, which prevents accumulation of Ci155, partially rescues proliferation defects in de2f1 mutant cells. Thus, the reduced expression of Ci155 in bel mutant cells appears to be important for the rescue of the de2f1 mutant phenotype. Since Ci155 is generally used as a marker of Hh pathway activity, it is possible that Hh signaling is reduced in bel mutant cells. This effect of bel is highly specific since other signaling pathways such as the Notch, Dpp, and EGFR pathways appear to be unaffected. It is worth noting that bel was independently isolated in a screen for genes required for photoreceptor differentiation, which also identified a number of Hh pathway components (24). However, the loss of bel does not completely phenocopy the loss of Hh signaling. In particular, the entry into the SMW that is indirectly dependent on Hh signaling (2, 15, 17) is relatively normal in bel mutant cells. One possible explanation is that the residual Ci155 present in bel mutant cells might be sufficient to promote the SMW but insufficient for cell cycle arrest in the MF. Interestingly, bel has been identified in an initial genome-wide RNAi screen as a potential candidate involved in the Hh signaling pathway (38). Our analysis of the bel mutant phenotype adds new significance to this finding, and we further expand these initial studies by examining the consequences of bel depletion and bel overexpression on an Hh reporter in a different cell line. However, although our data support the idea that the loss of bel rescues the de2f1 mutant phenotype at least partially through down-regulation of Ci155, the inability of the smo mutant allele to fully suppress the de2f1 mutant phenotype, to the extent of bel mutations, implies the existence of another, currently unidentified mechanism by which bel rescues the de2f1 mutant phenotype.

Based on sequence homology and its ability to functionally substitute for the yeast protein Ded1p, Bel has been assigned to the Ded1p subfamily of DEAD box proteins (26). Although Ded1p was implicated in the regulation of pre-mRNA splicing and mRNA export, the mostly well-defined role of Ded1p is in translational initiation (8, 10, 30). We disfavor the explanation that the loss of bel leads to a general reduction in protein synthesis because Minute mutants, which are characterized by a reduction in protein translation, proliferate more slowly than bel mutant cells. Additionally, mutations in eIF4A, a DEAD box protein that acts as a subunit of eIF4F in translational initiation, inhibits DNA replication (21). Since bel mutant cells proliferate normally and the loss of bel actually rescues the proliferation block of de2f1 mutant cells, it is unlikely that bel mutations reduce the capacity of the cell to translate proteins. Interestingly, mutations in ded1, the Schizosaccharomyces pombe homolog of bel, specifically affect translation of mRNAs encoding Cig2 and Cdc13 (22). Ded1 has been isolated in a multicopy screen of a cold-sensitive mutant of the fission yeast Cdc2 and interacted with checkpoint kinase Chk1 (32). This raises the possibility that Bel may also be involved in translational regulation of specific mRNAs, for example, those encoding components or regulators of the Hh signaling pathway. Such a model is supported by the findings that Bel is almost exclusively localized to the cytoplasm. Alternatively, Bel may exert its effect independently of regulation of translation. It has been recently shown that eIF4A inhibits Dpp signaling by promoting ubiquitination and degradation of two downstream targets of Dpp in Drosophila, Mad and Medea (28). Thus, DEAD box proteins may have diverse and unexpected functions in the cell.

The genetic rescue of the de2f1 mutant phenotype by the loss of bel underscores the potency of mutation of bel in overriding the cell cycle arrest in de2f1 mutant cells. These findings closely parallel studies of mammalian cells since DDX3, the mammalian homolog of Bel, has been recently implicated in negative regulation of cell proliferation. Intriguingly, DDX3 is down-regulated at a high frequency in hepatocellular carcinoma patients, suggesting that DDX3 might be a candidate tumor suppressor (6, 7). Given that Hh signaling is reduced in bel mutant cells, it would be interesting to determine if DDX3 also affects this developmental pathway in mammals. In conclusion, the results described here illustrate the advantages of a genetic approach in studying the de2f1 mutant phenotype, which may provide novel insights into the in vivo mechanisms of cell cycle regulation by dE2F and RBF.