The growth suppressive function of the retinoblastoma (pRB) tumor suppressor family is largely attributed to its ability to negatively regulate the family of E2F transcriptional factors and, as a result, to repress E2F-dependent transcription. Deregulation of the pRB pathway is thought to be an obligatory event in most types of cancers. The large number of mammalian E2F proteins is one of the major obstacles that complicate their genetic analysis. In Drosophila, the E2F family consists of only two members. They are classified as an activator (dE2F1) and a repressor (dE2F2). It has been previously shown that proliferation of de2f1 mutant cells is severely reduced due to unchecked activity of the repressor dE2F2 in these cells. We report here a mosaic screen utilizing the de2f1 mutant phenotype to identify suppressors that overcome the dE2F2/RBF-dependent proliferation block. We have isolated l(3)mbt and B52, which are known to be required for dE2F2 function, as well as genes that were not previously linked to the E2F/pRB pathway such as Doa, gfzf, and CG31133. Inactivation of gfzf, Doa, or CG31133 does not relieve repression by dE2F2. We have shown that gfzf and CG31133 potentiate E2F-dependent activation and synergize with inactivation of RBF, suggesting that they may act in parallel to dE2F. Thus, our results demonstrate the efficacy of the described screening strategy for studying regulation of the dE2F/RBF pathway in vivo.
THE family of E2F transcription factors and retinoblastoma (pRB) family tumor suppressor proteins play a pivotal role in control of cell proliferation (reviewed in Cayirlioglu and Duronio 2001; Trimarchi and Lees 2002; Blais and Dynlacht 2004; Dimova and Dyson 2005; Degregori and Johnson 2006). Although E2F is involved in a variety of cellular activities, the best understood function of E2F is to regulate transcription of genes at the G1/S transition. Among E2F targets are genes that encode regulators of S-phase entry and components of the DNA replication machinery. In mammals, there are eight E2F genes. E2F-1 through E2F-6 function as heterodimers with a DP subunit, while E2F-7 and E2F-8 do not require DP to bind to DNA. In spite of structural similarities among E2F proteins, E2F-1, E2F-2, and E2F-3a are predominately involved in activation of gene expression while the group E2F-3b, E2F-4, E2F-5, E2F-6, E2F-7, and E2F-8 behave as repressors. In quiescent cells, when activity of cyclin-dependent kinases (cdks) is low, repressor E2Fs are complexed with the pRB family members (also called pocket proteins) and repress expression of E2F regulated genes. The prevailing mechanism of the repression is thought to be through the direct recruitment of histone deacetylases, histone methylases, and other corepressor complexes by pocket proteins to E2F regulated promoters. Upon entry into the cell cycle, mitogenic stimulation leads to an increase in the activity of G1 cdks, which phosphorylate pRB family members and disrupt their interaction with E2Fs. This coincides with displacement of the repressor E2Fs, appearance of free activator E2Fs on the promoters of E2F responsive genes, and induction of the E2F transcriptional program. The critical role of the E2F/pRB network in cell proliferation is underscored by obligatory inactivation of pRB control in almost all cancers (Hanahan and Weinberg 2000). The prevailing view is that the tumor suppressor property of pRB is to constrain E2F activity. However, our knowledge of the various tiers of regulation by which pRB has the capacity to block cell proliferation in the context of a multicellular organism is still very limited. Thus, understanding how the growth suppressive function of pRB/E2F can be overridden is important to fully understand the importance of the pRB pathway in cancer and during normal development.
Drosophila represents an ideal model system to address this question. This is primarily due to the ability to carry out high-throughput genetic screens to unveil novel functional interactors of the E2F/pRB pathway in an unbiased way. Importantly, the pRB/E2F network is highly conserved in Drosophila, yet the families are much smaller than in mice and humans. The Drosophila genome encodes an activator, dE2F1, and a repressor, dE2F2, their heterodimeric partner, dDP, and two pocket proteins, RBF1 and RBF2 (reviewed in Cayirlioglu and Duronio 2001; Dimova and Dyson 2005; van den Heuvel and Dyson 2008). Studies of the phenotypes of de2f and rbf mutant and transgenic animals have provided clear insights into the critical roles of these genes in cell proliferation (Duronio et al. 1995, 1998; Royzman et al. 1997; Neufeld et al. 1998; Du and Dyson 1999; Cayirlioglu et al. 2001; Frolov et al. 2001). As in mammals, dE2F1 is a potent inducer of S-phase entry. Overexpression of de2f1 drives postmitotic cells of the eye imaginal disc into the cell cycle (Asano et al. 1996; Brook et al. 1996). Consistently, inactivation of rbf1, which negatively regulates de2f1, leads to ectopic cell cycles (Du and Dyson 1999; Firth and Baker 2005). Analysis of single and double de2f mutant animals revealed that de2f1 and de2f2 act antagonistically during larval development. This conclusion is based on the observation that the phenotype of de2f1 de2f2 double mutant animals is less severe than the phenotype of de2f1 mutants (Frolov et al. 2001). Cell proliferation and E2F-dependent transcription are severely reduced in de2f1 mutant animals. These defects are suppressed by a concomitant mutation of de2f2. The pattern of cell proliferation is largely normal and repression of several examined E2F targets is relieved in de2f1 de2f2 double mutants. These results indicate that unchecked de2f2 provides a significant contribution to the de2f1 mutant phenotype and therefore, the lack of cell proliferation in de2f1 mutants can be potentially viewed as a readout of the de2f2 activity.
Here, we present the results of a mosaic genetic screen aimed at identifying suppressors that rescue proliferation in de2f1 mutants [Su(E1)'s]. Since the de2f1 mutant phenotype stems from the cell proliferation block by dE2F2/RBF, we expected that some of the isolated suppressors would be important for dE2F2/RBF function. Indeed, we have identified mutations in l(3)mbt and B52, two genes that were previously implicated in dE2F2/RBF-mediated repression. We have also identified two suppressors, gfzf and CG31133, which potentiate E2F-dependent transcription and cooperate with the dE2F/RBF pathway. Thus, the characterization of Su(E1)'s can provide new insights into the in vivo regulation of the dE2F/RBF pathway.
MATERIALS AND METHODS
Drosophila were cultured on standard media at 25°. A description of all alleles and fly stocks can be found in FlyBase.
For mutagenesis, 36- to 72-hr-old w; FRT 82B de2f1729/TM3 Sb males were starved for 3–6 hr and fed 25 mm EMS in a 1% sugar solution for 16–18 hr. In the F1 screen, mutagenized males were crossed to ey-FLP; FRT 82B l(3)cl-R31/TM6B females and the progeny were screened for exceptional males that contained visible clones of de2f1 mutant cells. In the F2 screen, mutagenized males were first crossed to w; TM3 Ser/MKRS females to balance putative suppressors of the de2f1 mutant phenotype. Then the individual FRT82B Su(E1) de2f1729/TM3 or FRT82B Su(E1) de2f1729/MKRS males were crossed to three ey-FLP; FRT 82B l(3)cl-R31/TM6B females to identify potential suppressors of the de2f1 mutant phenotype. Isolated Su(E1)'s were retested multiple times by crossing them to ey-FLP; FRT 82B l(3)cl-R31/TM6B and ey-FLP; FRT 82B P[Ubi-GFP]/TM6B females.
To map isolated Su(E1)'s, females that were w; FRT82B de2f1729 Su(E1)/TM3 Ser were mated to males heterozygous for the 3R chromosomal deficiencies from the Exelixis and DrosDel deficiency kits. The following deficiencies that uncover gaps present in the Exelixis and DrosDel kits were also included in complementation tests: Df(3R)Antp-X1, Df(3R)BSC24, Df(3R)by10, Df(3R)sbd26, Df(3R)P2, Df(3R)DG2, Df(3R)Cha1a, Df(3R)Dl-BX12, Df(3R)BSC43, Df(3R)BSC56, Df(3R)Espl3, Df(3R)BSC42, Df(3R)L127, and Df(3R)B81. A recovery of <1% of adult flies trans-heterozygous for Su(E1) and a deficiency relative to their heterozygous siblings were considered a failure to complement. Since Su(E1)'s were isolated in the presence of the de2f1729 mutation, each Su(E1) was lethal in trans to Df(3R)Exel6186, which uncovers the de2f1 gene.
To map Su(E1)-A the exon–intron junctions and ORFs of the following genes were sequenced: CG18682, CG11498, CG1359, CG2171, CG31025, CG9753, CG31030, CG1983, CG9747, G31029, CG15529, CG31028, CG15530, and CG15531. Genomic DNA was extracted from adult flies of w; FRT82B de2f1rm729 Su(E1)-A7d13/TM6B Tb and sequenced. No changes resulting in premature stop codons or affecting exon–intron junctions were found. However, we found single-nucleotide changes that do not result in the above-mentioned defects in the sequence of the following genes: CG18682, CG11498, CG31025, CG15530, CG15531, CG31028, CG31029, CG9747, and CG31030. To determine whether any of these single-nucleotide changes were newly induced mutations or single-nucleotide polymorphisms, genomic DNA was isolated from the following genotypes and used as a control in sequencing: w; FRT82B de2f1rm729 Su(E1)5d7/TM6B Tb and w; FRT82B de2f1rm729 Su(E1)7d21/TM6,Tb. In several cases the regions of interest were sequenced using genomic DNA of the following genotypes: w; FRT82B de2f1rm729 Su(E1-A)7d13/TM3 Ser and w; FRT82B de2f1rm729 Su(E1-A)7d13/TM6 Tb [ubi-GFP]. Additionally, we sequenced several genes using genomic DNA isolated from flies carrying another allele of Su(E1)-A6a39. On the basis of these comparisons we concluded that all single-nucleotide changes that we found are SNPs of the parental chromosome and therefore are not newly induced mutations in the above-mentioned genes.
For Su(E1)'s, Su(E1)-A7d13, CG311337d21, and gfzf6a27, an FRT82B Su(E1) de2f1+ chromosome was generated from the FRT82B Su(E1) de2f1729 chromosome by precise excision of the P element from the de2f1729 allele. Since both the de2f1729 allele and FRT82B are marked with the rosy gene, excisions of the P element from the de2f1729 allele could not be followed by the rosy marker. Therefore, the FRT82B Su(E1) de2f1729/Δ2–3 Sb males were crossed to TM3 Ser/MKRS females and ∼100 rosy+ males were crossed individually to de2f17172/TM3 Sb females to select chromosomes in which the de2f1729 allele was reverted to wild type. The presence of a functional FRT82B was verified by crossing to ey-FLP; FRT82B P[Ubi-GFP, mini-white] to visualize the appearance of clones.
Eye and wing imaginal discs were dissected from third instar larvae in Schneider's insect medium (GIBCO, Grand Island, NY) and then fixed in phosphate-buffered saline (PBS) buffer with 4% formaldehyde. Washes were then done in PBS, 0.3% Triton X-100. Discs were then blocked in PBS, 0.1% Triton X-100, 10% normal donkey serum for 1 hr followed by the addition of primary antibodies. Discs were incubated overnight with primary antibodies and then washed in PBS, 0.1% Triton X-100 three times for 10 min each. Discs were then incubated in blocking solution containing the appropriate secondary antibodies. Washes were then done using PBS, 0.1% Triton X-100. Discs were then placed into glycerol containing 0.5% propyl gallate in preparation for slide mounting. Bromodeoxyuridine (BrdU) (Sigma, St. Louis) labeling of eye imaginal discs was performed by dissecting discs and incubating them in 0.3 mg/ml BrdU–Schneider's insect medium for 30 min. Washes were then done in PBS followed by an overnight fixation in 1.5% formaldehyde. Washes were then done in PBS before incubating discs with DNAse [Promega (Madison, WI) RQ1] for 45 min in a 37° water bath. Then washes, blocking, and antibody incubations followed as described above. Images were collected with a Zeiss LSM510 confocal microscope. The primary antibodies used in this work were mouse anti-BrdU (1:50) (BD Bioscience), rat anti-Elav (1:50) (Developmental Studies Hybridoma Bank), and rabbit anti-phospho-H3 (1:150) (Upstate). The secondary conjugated antibodies used in this work were anti-mouse Cy3 (1:100), anti-rabbit Cy5 (1:100), and anti-rat Cy3 (1:50) from Jackson Immuno Laboratories.
RNAi, Western blot analysis, transient transfections, and real-time PCR in Drosophila tissue culture cells:
For RNAi treatments ∼2.5 × 104 S2R+ cells per well were plated in a 24-well plate containing 400 μl of Schnieder's insect medium with 10% fetal bovine serum (Sch + FBS). (Schnieder's insect medium was from GIBCO and FBS was from Atlanta Biologicals). Cells were allowed to adhere to the plate for 1 hr before being washed with medium. Then 12.5 μg of each appropriate dsRNA in 200 μl medium were added to each appropriate well. After 3 hr, 400 μl of Sch + FBS were added to each well. After 4 days, appropriate wells were treated with dE2F1 dsRNA following the above procedure for 18 hr. Then 200 μl of cells per well were replated in duplicate in a 24-well plate containing 200 μl Sch + FBS. Cells were allowed to adhere for 1 hr before a transfection solution consisting of appropriate DNA and a transfection agent, FuGENE HD (Roche Diagnostics, Indianapolis), in 200 μl Sch was added to each well. After ∼30 hr cells were lysed in 1 mm EDTA, 100 mm NaCl, 10 mm Tris-HCl (pH 7.8), and 0.25% NP-40 before assaying for luciferase and β-galactosidase activity. Luciferase assays were performed on a BIO-TEK Clarity Microplate Luminometer. β-Galactosidase assays were performed on a Labsystems Multiskan MS Plate Reader. Quantitative real-time PCR on RNA isolated from S2 cells and primers were as previously described (Rasheva et al. 2006). For Western blot analysis, 5–10 × 106 cells were lysed in 250 μl of 100 mm NaCl, 10 mm tris-HCl, 1 mm EDTA, and 0.25% NP-40 (pH 7.8) buffer and frozen for 5 min at −80°. Proteins were resolved by sodium dodecyl sulfate–10% polyacrylamide gel electrophoresis. Guinea pig polyclonal anti-dE2F1 (Bosco et al. 2001) and mouse anti-tubulin antibodies were used for Western blot analysis.
A suppressor screen for genes that relieve the de2f2-dependent block to cell proliferation:
Genetic mosaic screens based on mitotic recombination have proved to be particularly effective in identifying genes that regulate cell proliferation. The ey-FLP/FRT technique provides a high level of mitotic recombination, thus allowing for the generation of homozygous mutant tissue in the eye with high efficiency. Since the wild-type chromosome carries the white gene (P[mini-white]), the homozygous mutant tissue can be easily distinguished by the lack of the white eye color marker in the adult eye (Figure 1A). Thus, the ey-FLP/FRT-based screen is an attractive approach for identifying novel genes that are important for dE2F2/RBF function in a high-throughput manner. Additionally, we sought a genetic strategy that meets the following criteria: does not rely on overexpression of de2f or rbf genes and is sufficiently sensitive. The de2f1 mutant phenotype meets these criteria. Clones of de2f1 mutant cells are extremely small due to a strong proliferation block in these cells (Brook et al. 1996; Neufeld et al. 1998). Since proliferation of de2f1 mutant cells can be rescued by inactivation of de2f2, patches of de2f1 de2f2 double mutant tissue can be found in the adult eye (Ambrus et al. 2007). An example of a clone in which both de2f1 and de2f2 are functionally inactivated by a dDP mutation is shown in Figure 1B. Therefore, we expected that mutations that compromise the dE2F2/RBF function should act as suppressors of the de2f1 mutant phenotype [Su(E1)'s] and could be identified by the appearance of clones of de2f1 Su(E1) double mutant cells in the adult eye.
To widen the observable range of suppression of the de2f1 mutant phenotype, we adopted an ey-FLP/FRT cell lethal system in which most of the homozygous wild-type tissue is eliminated due to the presence of a recessive cell lethal mutation on a wild-type chromosome (Newsome et al. 2000). When clones of E2F-deficient cells were induced with the ey-FLP cell lethal technique using a dDP mutation, large patches of the mutant tissue could be easily identified in all animals (Figure 1, C and D). Importantly, even with the ey-FLP/FRT cell lethal technique, no white tissue, representing de2f1 mutant cells, can be seen in the eye (Figure 1E). Thus, the small size of clones of de2f1 mutant cells is a highly consistent phenotype, which is sensitive to inactivation of de2f2. We concluded that the de2f1 mutant phenotype might be suitable for a suppressor mosaic genetic screen to identify novel genes that affect the dE2F2/RBF function.
To test our screening strategy we initially performed an F1 screen. Isogenic FRT82B de2f1729/TM3 males were mutagenized with EMS and crossed to ey-FLP; FRT 82B l(3)cl-R31/TM6B females. The male F1 progeny of these flies were scored for the appearance of white patches indicating the presence of a potential Su(E1), which allows for the recovery of de2f1 mutant tissue. Such exceptional males were crossed to females carrying the third chromosomal balancers and then the male progeny were backcrossed to the ey-FLP; FRT 82B l(3)cl-R31/TM6B females to confirm the suppression of the de2f1 mutant phenotype. A total of 18,000 chromosomes were screened, and 75 F1 males having white patches of the de2f1 Su(E1) double mutant tissue were selected. However, only 8 mutants were successfully retested (Table 1). To improve the recovery of suppressors, we performed an F2 screen in which mutagenized FRT82B de2f1729 chromosomes were first balanced. Then FRT82B de2f1729/TM3 or FRT82B de2f1729/MKRS males carrying a potential Su(E1) were crossed individually to the ey-FLP; FRT 82B l(3)cl-R31/TM6B females (Figure 2). Of 44,600 individual crosses ∼40,000 crosses produced progeny and 244 mutant lines exhibited a suppression of the de2f1 mutant phenotype (examples are shown in Figure 1, F–J). Males of each of the 244 lines were backcrossed twice to females carrying the balancer chromosomes to replace the ey-FLP-carrying X chromosome with a wild-type X chromosome. Additionally, these males were retested to confirm consistency of the suppression of the de2f1 mutant phenotype.
To determine whether the five Su(E1)'s shown in Figure 1, F–J, are able to rescue the de2f1 mutant phenotype in tissues other than the eye, we induced clones of de2f1 Su(E1) double mutant cells in the third instar larval wing imaginal disc, using Ubx-FLP. The Ubx-FLP/FRT technique provides a high level of mitotic recombination in multiple tissues. In the wing imaginal disc, the homozygous wild-type tissue can be easily distinguished from homozygous mutant tissue by the presence of the GFP signal (Figure 1K). As expected, patches of clones of de2f1 mutant cells are limited to approximately one to five cells and therefore very little de2f1 homozygous mutant tissue (GFP negative) is found in the wing disc (Figure 1L). In contrast, much larger clones of de2f1 Su(E1) double mutant cells can be recovered (Figure 1, M–Q). Thus, examined Su(E1)'s are able to at least partially rescue the strong proliferation defects of de2f1 mutant cells in multiple tissues.
Initial classification of suppressors:
Next, we tested whether the suppressors rescue the proliferation defects of de2f1 mutant cells without the aid of the cell lethal mutation, l(3)cl-R31, on the wild-type chromosome. The ability of isolated mutations to rescue cell proliferation of de2f1 mutant cells in these settings provided a rigorous test of the strength of suppression since both wild-type and the mutant cells compete against each other in equal conditions. All of the 252 Su(E1)'s that were isolated in the F1 and F2 screens were crossed to ey-FLP; FRT82B P[Ubi-GFP; w+] females and the eyes of the progeny were examined for the appearance of clones of mutant cells. A total of 166 mutants were characterized as weak suppressors while 86 suppressors exerted a robust effect on the de2f1 mutant phenotype giving rise to visible white patches of de2f1 mutant tissue in the eye (Table 2). Among these 86 lines, five mutants produced a very strong rough eye phenotype with occasional scars indicating that these suppressors are likely to affect developmental pathways and therefore were not considered further.
Mapping and complementation analysis:
We used a lethality potentially associated with Su(E1)'s in an initial complementation analysis. Although the lethal mutations are not necessarily in each case Su(E1)'s that are responsible for the phenotype, such an approach allows for quick mapping of lethal Su(E1)'s. However, the presence of a lethal de2f1 mutation on each of the Su(E1)-containing chromosomes complicates complementation and mapping analyses. Each Su(E1) is lethal in trans to each other due to the presence of the de2f1729 mutation. Similarly, this precludes a standard mapping by meiotic recombination onto a chromosome with multiple recessive markers, since the presence of a Su(E1) can be determined only in the background of the de2f1 mutation. As an alternative way to estimate how many genes were represented by the 81 strongest Su(E1)'s, we tested these lines in trans to a combined Exelixis and DrosDel deficiency kit that uncovers the right arm of chromosome 3 (for details see materials and methods). Forty-nine lines fully complemented the deficiency kit, indicating that either corresponding Su(E1)'s are viable or they are located within the limited intervals between the breakpoints of the deficiencies. Each of the remaining 32 mutants were lethal in trans to a particular deficiency (Table 2). To determine whether the group of weak mutants contained additional alleles, we crossed weak mutants to identified deficiencies. In this way, 7 additional mutant alleles were identified. Publicly available known mutations and P-element insertions in the relevant regions uncovered by the deficiencies were tested for complementation to determine the identities of Su(E1)'s. This allowed us to identify new alleles of l(3)mbt, Doa, gfzf, B52, and CG31133 as suppressors of the de2f1 mutant phenotype. The mapping results are summarized in Table 2 and described below.
Seven suppressors (6a14, 4d30, 3d23, 5d8, 4d16, 5d6, 6a31, and 3d33) were mapped to the interval between 97E2 and 98A7. Each of these alleles was lethal in trans to the deficiency Df(3R)ED6265. One of the genes uncovered by the deficiency is l(3)mbt, which encodes a protein associated with the dE2F2/RBF repressor complex DREAM/MMB (Lewis et al. 2004). Each of the seven suppressors was lethal in trans to the nonsense allele l(3)mbtE2 and to the P-element insertion in l(3)mbt, f02565. Additionally, all seven were lethal in trans to a deficiency, Df(3R)D605, which uncovers l(3)mbt. This suggests that this group of Su(E1)'s is allelic to l(3)mbt.
Another group included five Su(E1)'s: 6d2, 6a21, 5d19, 6a52, and 5d7. Each mutant of this group failed to complement Df(3R)Exel6210. Testing for complementation with known mutations and P-element insertions in these regions identified an allele of the Darkener of apricot (Doa) gene, Doa01705b, which failed to complement each of five Su(E1)'s. Similar results were obtained with two other Doa alleles, DoaKG09056 and Doa3. This suggests that this group of Su(E1)'s consists of mutations in the Doa gene. Doa mutations were initially isolated as suppressors of the weak white-apricot allele (Rabinow et al. 1993).
The B52 gene encodes an SR protein that is required for the correct splicing of the de2f2 pre-mRNA. Accordingly, mutations in B52 were shown to rescue proliferation of de2f1 mutant cells (Rasheva et al. 2006). Two new mutations in the B52 gene, 3d16 and 3d25, were isolated. Both 3d16 and 3d25 failed to complement the deficiency uncovering the B52 gene Df(3R)Exel6169 and the B52s2249 mutant allele.
A Su(E1), 7d21, was mapped to the region 95E5–F8, which is uncovered by the deficiencies Df(3R)Exel6198 and Df(3R)ED6187. To further narrow down the position of 7d21, we generated two smaller deletions, Df(3R)7d21.I and Df(3R)7d21.II, within this interval according to the method described in Parks et al. (2004). In this technique, a deletion is generated between two FRT-bearing insertions in the presence of the FLP recombinase. We utilized a pair of XP insertions, d01966 and d05015, to generate Df(3R)7d21.I while Df(3R)7d21.II was recovered using a pair of XP insertions, d09860 and d05397. 7d21 was found to be lethal in trans to Df(3R)7d21.I, but not to Df(3R)7d21.II. Among publicly available mutations in genes uncovered by Df(3R)7d21.I only insertion f07858 failed to complement 7d21. f07858 is an insertion in the gene CG31133, which encodes a protein of unknown function. To unambiguously confirm that a mutation in CG31133 is responsible for the rescue of the de2f1 mutant phenotype, we recombined CG31133f07858 with the de2f1729 mutant allele and found that clones of de2f1729 CG31133f07858 double mutant cells can be recovered in the eye (data not shown). This suggests that CG31133 is a novel suppressor of the de2f1 mutant phenotype.
GST-containing FLYWCH zinc-finger protein:
A lone Su(E1) 6a27 failed to complement two deficiencies Df(3R)ED5221 and Df(3R)ED7665, which uncover the interval 84C4–E10. Candidate mutations in this interval were tested for their ability to complement 6a27. 6a27 was lethal in trans to an EP insertion EY08448. EY0848 is inserted into the GST-containing FLYWCH zinc-finger protein (gfzf) gene, which is located within a second intron of CG2656. Thus, EY08448 disrupts both gfzf and CG2656. To determine which of the two genes is affected by 6a27, we used the KG08427 and e00303 P-element insertions in a complementation test with 6a27. In KG08427 and in e00303, the P element is inserted into the first and the second exons of CG2656, respectively. Hence, the P-element insertions are upstream of the gfzf gene and therefore, gfzf function is unlikely to be affected by either KG08427 or e00303. 6a27 was found to be fully viable in trans to KG08427 and e00303. In contrast, 6a27 failed to complement two gfzf mutant alleles, gfzf2 and gfzfcz811. gfzfcz811 contains a deletion at residue F469, producing a frameshift to a stop codon at residue 476 (Provost et al. 2006). Thus, 6a27 fails to exclusively complement mutations in the gfzf gene, but is fully viable in trans to mutant alleles of CG2656. These data suggest that 6a27 is a novel mutant allele of gfzf.
Two Su(E1)'s, 7d13 and 6a39, were lethal in trans to Df(3R)Exel6214, which deletes the region of 99D5–E2. According to FlyBase (Drosophila genome sequence R5.17), there are 17 predicted genes within this interval. We tested available mutant alleles of CG15531, CG7951, CG2184, CG31027, and CG9747 and found that each fully complements 7d13 and 6a39, indicating that these 5 genes do not correspond to Su(E1)-A. We have sequenced the ORFs and intron–exon junctions of the remaining 12 genes as well as CG15531 and CG9747 (for details see materials and methods). No newly induced mutations in these genes were found. There are a couple of possible explanations to account for this. First, lethality associated with Su(E1)-A might be unrelated to the suppressor effect. This seems unlikely given that we isolated two independent alleles, 7d13 and 6a39, which were mapped to the same genomic interval. Another explanation is that these alleles carry mutations in a regulatory region of a gene. Such mutations can potentially affect the level of transcription or translation but would be missed in our analysis. Finally, it is possible that there is another currently unannotated gene within this genomic interval that corresponds to Su(E1)-A.
For our initial analysis, we studied the phenotypes of five above-mentioned Su(E1)'s by examining markers of cell proliferation in clones of de2f1 Su(E1) double mutant cells in the eye imaginal disc since the pattern of cell proliferation is well characterized. In the larval eye imaginal disc, the pattern of cell divisions is defined by the position of the morphogenetic furrow (MF), which traverses from posterior to anterior during development. We labeled the eye disc with BrdU to visualize cells in S phase. In the wild-type disc, cells are asynchronously dividing anterior to the MF and then become transiently arrested in G1 within the MF (Figure 3A). Upon emerging from the G1 arrest in the MF, cells enter the last S phase in a highly synchronous manner to form a tight stripe of BrdU-positive cells called the second mitotic wave (SMW). Posterior to the SMW cells exit the cell cycle and commit to neuronal differentiation. Unlike wild-type cells, de2f1 mutant cells fail to enter the SMW (Du 2000). This defect is due to the presence of de2f2 since de2f1 de2f2 double mutant cells and dDP single mutant cells enter the SMW normally although with a slightly reduced rate of S-phase progression (Frolov et al. 2005; Ambrus et al. 2007). Therefore, we categorized Su(E1)'s by the extent to which they restore the SMW in de2f1 mutant cells. We chose to analyze 3d33, 6d2, 6a27, 7d21, and 7d13, which represent mutant alleles of l(3)mbt, Doa, gfzf, CG31133, and Su(E1)-A, respectively. Clones of double mutant cells carrying a Su(E1) and the de2f1729 allele were induced with ey-FLP. Homozygous mutant tissue was distinguished by the lack of GFP. From this analysis we found that Su(E1)'s restored the SMW in de2f1 mutant cells to varying extents. Mutant alleles of l(3)mbt and Su(E1)-A provided a stronger rescue of the SMW in de2f1 mutant cells as evident by the appearance of BrdU-positive de2f1 l(3)mbt3d33 and de2f1 Su(E1)-A7d13 double mutant cells in the SMW at about the same time as in the adjacent wild-type cells (Figure 3, B and C). However, we note that the intensity of BrdU labeling was somewhat reduced in de2f1 Su(E1)-A7d13 mutant cells. In contrast to l(3)mbt and Su(E1)-A, de2f1 CG311337d21 double mutant cells entered the SMW several columns more posterior than wild-type cells (Figure 3D), which is indicative of a delay of entry into the S phase. Additionally, BrdU incorporation persisted several columns more posterior in the double mutant cells compared to the adjacent wild-type tissue. This indicates that it takes longer than normal for de2f1 CG311337d21 double mutant cells to progress through the S phase, suggesting that the S phase is likely to be slowed. The defects in the SMW were less noticeable in the case of the gfzf6a27 mutant allele (Figure 3E). Although occasionally we observed clones of de2f1 gfzf6a27 double mutant cells in which entry into the SMW occurred relatively on time, in most cases, the BrdU-positive mutant cells were situated farther posterior. Finally, de2f1 Doa6d2 double mutant cells were only sporadically labeled with BrdU in the SMW. However, any de2f1 Doa6d2 double mutant cells that did incorporate BrdU were always found to be significantly more posterior than BrdU incorporating adjacent wild-type cells (Figure 3F). Taken together these results indicate that Su(E1)'s differ among each other by the extent of rescue of the timing of entry into and exit from the SMW in de2f1 mutant cells. Clones of de2f1 l(3)mbt and de2f1 Su(E1)-A mutant cells enter and exit the SMW relatively on time while de2f1 gfzf6a27, de2f1 CG311337d21, and de2f1 Doa6d2 exhibit a delay in the entry into and/or exit from the SMW.
The extent of the defects in the SMW may reflect the ability of each Su(E1) to rescue the cell cycle arrest in de2f1 mutant cells. Alternatively, Su(E1) mutant cells themselves could have an aberrant SMW. To discriminate between these two possibilities we examined BrdU incorporation in the SMW of clones of Su(E1)-A, CG31133, gfzf, and Doa single mutant cells. As shown in Figure 3, G–J, Su(E1)-A, CG31133, or gfzf single mutant cells enter and exit the SMW at about the same time as the adjacent wild-type cells, while the SMW entry is slightly delayed in Doa single mutant cells. However, this delay is not as extreme as that observed in de2f1 Doa double mutant cells (Figure 3F). We infer from these results that the defects in the SMW of clones of de2f1729 Su(E1) mutant cells likely reflect the ability of each Su(E1) to rescue the SMW in de2f1729 mutant cells.
To further characterize the effect of Su(E1)'s on the SMW in de2f1 mutant cells we used phospho-H3 as a mitotic marker (Figure 4). In wild-type eye discs, the stripe of phospho-H3-positive cells posterior to the MF marks cells of the SMW (Figure 4A). Importantly, we found phospho-H3-positive double mutant and single mutant cells in the SMW (Figure 4, B–I). Expression of ELAV, a marker of neuronal differentiation, is induced in double and single mutant cells although de2f1 Doa double mutant and Doa single mutant cells exhibited a slight delay in the onset of ELAV expression (Figure 5 ). This could be a consequence of a delay in exit from the SMW. In no cases did we find cells that coexpress phospho-H3 and ELAV. This suggests that de2f1 Su(E1) double mutants finish the SMW and commit to differentiation.
l(3)mbt affects expression of an E2F reporter in cell culture-based assays differently than gfzf or CG31133:
Having established that the loss of l(3)mbt restores the timing of the S-phase entry in the SMW in de2f1 mutant cells, while mutations in gfzf and CG31133 do not, we wished to determine whether inactivation of these genes differentially affects E2F-dependent transcription in dE2F1-deficient cells. We employed RNA interference (RNAi) in Drosophila S2R+ tissue culture cells to deplete dE2F1, dE2F2, RBF, L(3) malignant brain tumor (MBT), GFZF, and CG31133 proteins and then transiently transfected these cells with a reporter containing the endogenous PCNA promoter fused to a luciferase gene. The PCNA gene is a direct transcriptional target of dE2F and has been previously validated as an accurate readout of dE2F transcriptional activity (Frolov et al. 2001). As expected, the E2F reporter is strongly repressed in dE2F1-depleted cells (Figure 6A). This is due to the presence of dE2F2 because the repression was alleviated when dE2F1 and dE2F2 were inactivated simultaneously. Accordingly, depletion of RBF proteins, which are required for repression by dE2F2, also relieved repression of the reporter (Figure 6A). Thus, expression of the E2F reporter in dE2F1 single- and in dE2F1 dE2F2 double-depleted cells essentially recapitulates the response of the endogenous PCNA gene in mutant animals or in tissue culture cells (Frolov et al. 2005). Inactivation of GFZF or CG31133 had no effect on the E2F reporter in dE2F1-deficient cells, as evidenced by the reporter remaining fully repressed in these cells (Figure 6A). In contrast, depletion of L(3)MBT relieved the repression of the reporter in dE2F1-deficient cells to the same extent as inactivation of dE2F2 (Figure 6A). To test whether repression of an endogenous PCNA gene in dE2F1-depleted cells is alleviated by codepletion of L(3)MBT, we used real-time RT–PCR to monitor the steady-state PCNA mRNA level. As has been previously shown, PCNA expression is reduced in dE2F1-depleted cells (Frolov et al. 2005). Consistent with the results in transient transfections, the repression of PCNA was relieved in dE2F1 L(3)MBT double-deficient cells to a level comparable to that of dE2F1 dE2F2-depleted cells (Figure 6B). Thus, inactivation of L(3)MBT alleviates the repression of an E2F target in dE2F1-depleted cells as efficiently as removal of dE2F2, indicating that L(3)MBT is needed for dE2F2 to repress.
Next, we determined the effect of depletion of L(3)MBT, CG3133, GFZF, and DOA on the basal level of expression of the E2F reporter. Interestingly, although depletion of GFZF or CG31133 does not relieve dE2F2-mediated repression, we found that the E2F reporter is consistently elevated in cells lacking GFZF or CG31133 (Figure 6C). This effect cannot be attributed to an elevation of dE2F1 protein since dE2F1 levels are unaltered when any of the Su(E1)'s are depleted in cell culture by RNAi (Figure 6D). To determine whether these effects are mediated through pocket proteins we determined the epistatic effect of depletion of GFZF and CG31133 relative to RBF proteins. As expected, depletion of RBF proteins elevates expression of the E2F reporter (Figure 6E), which is likely due to release of an inhibitory effect of RBFs on endogenous dE2F1. Codepletion of L(3)MBT does not potentiate the effect of inactivation of RBF proteins. This is consistent with the finding that L(3)MBT and RBF are present together in the same repressor complex (Lewis et al. 2004). Unlike L(3)MBT, codepletion of GFZF and RBFs or CG31133 and RBFs resulted in strong additive effects on activation of the E2F reporter as compared to targeting either of them alone (Figure 6E). These observations support the idea that inactivation of GFZF or CG31133 elevates activation of the E2F reporter, but this occurs in an RBF-independent manner.
Although a critical role of the pRB/E2F pathway in normal cell proliferation and in cancer has been firmly established, our understandings of what aspects of pRB/E2F function are important in vivo and how precisely the pathway is regulated during development are still very limited. As a way to begin to address these important questions, we have designed a mosaic genetic screen in Drosophila to isolate genes that when mutated overcome the dE2F2/RBF-dependent block to cell proliferation in de2f1 mutant cells. Several genetic modifier screens have been previously performed in Drosophila to dissect the pRB/E2F pathway (for examples see Staehling-Hampton et al. 1999; Lane et al. 2000; Weng et al. 2003). These screens provided critical contributions to the understanding of mechanisms underlying actions of pRB and E2F. However, such screens were designed to exclusively isolate haploinsufficient modifiers in the background of ectopically expressed rbf, de2f, and cyclin E genes, thus limiting the spectrum of potential modifiers. The screening strategy described in this work avoids complications associated with protein overexpression since the cell cycle block in de2f1 mutant cells is caused by the endogenously present de2f2. Additionally, the FLP/FRT technique allows for the isolation of recessive suppressors. These are the two major advantages of our approach from the above-mentioned screens.
The genetic screen reported here identified genes that are known to be required for dE2F2 function as well as novel genes that were not previously linked to the pRB/E2F pathway. Of the complementation groups isolated in the screen, l(3)mbt and B52 have been previously implicated in dE2F2-mediated repression (Lewis et al. 2004; Rasheva et al. 2006; Lu et al. 2007). L(3)MBT has been biochemically purified as a component of a native dE2F/RBF complex Drosophila RBF E2F and MYB/Myb-MuvB (DREAM/MMB), although in a substoichimetric ratio (Lewis et al. 2004). Mutations in l(3)mbt give rise to malignant transformations in the larval brain as well as disruption of synchronous cell divisions in early embryos (Yohn et al. 2003; Trojer and Reinberg 2008). l(3)mbt encodes a protein with three MBT domains that are structurally similar to other chromatin-binding domains such as the chromodomain. Its mammalian homolog L3MBTL1 is a transcriptional repressor that is thought to promote the higher-order chromatin structure by simultaneously binding to two adjacent nucleosomes and moving nucleosomes closer together. Such binding is dependent on the presence of mono- and dimethylation of histone H4 at lysine 20, which is commonly found in facultative heterochromatin (Trojer et al. 2007). L3MBTL1 associates with pRB and is needed for repression of E2F-regulated genes c-myc and cyclin E in mammalian cells (Trojer et al. 2007). Interestingly, depletion of L(3)MBT in Drosophila tissue culture cells results in derepression of genes that are normally kept silent by dE2F2/RBF in asynchronously dividing cells (Lewis et al. 2004; Lu et al. 2007). In this respect, isolation of l(3)mbt as a suppressor of the de2f1 mutant phenotype described here is significant because it provides the first genetic evidence for the importance of l(3)mbt in de2f2-dependent cell cycle arrest in vivo. This conclusion is further supported by the finding that L(3)MBT is needed for dE2F2 to repress in dE2F1-deficient cells. Finally, isolation of l(3)mbt serves as a proof of principle that genes that interface directly with the dE2F/RBF module can be identified as suppressors of the de2f1 mutant phenotype.
In addition to l(3)mbt, three genes, rbf2, Chromatin assembly factor 1 subunit (Caf1/p55), and rpd3, reside on 3R and encode components of the DREAM/MMB complex. However, rpd3 is not required for the repression of dE2F2/RBF-regulated genes (Korenjak et al. 2004; Lewis et al. 2004; Taylor-Harding et al. 2004; Georlette et al. 2007) while rbf2 is primarily involved in E2F-dependent repression in embryos, but not in imaginal discs (Stevaux et al. 2005), and therefore it is not surprising that mutant alleles of rpd3 and rbf2 were not isolated in this screen. The lack of Caf1/p55 mutations among Su(E1)'s is somewhat puzzling given that inactivation of CAF1/p55 abrogates dE2F2-mediated repression (Taylor-Harding et al. 2004). One possibility is that the CAF1/p55 function itself is needed during cell proliferation due to its role in DNA replication. Alternatively, since no mutant alleles in Caf1/p55 have been described to date it is not known whether mutations in Caf1/p55 are lethal. If the loss of Caf1/p55 does not result in lethality, then mutant alleles of Caf1/p55 would be apparently missed in our initial mapping approach using lethality as a complementation test.
Among Su(E1)'s we have isolated multiple mutant alleles of the Doa gene. Doa has been implicated in a variety of cellular functions including differentiation and cell cycle progression (Yun et al. 2000; Bettencourt-Dias et al. 2004; Bjorklund et al. 2006). The Doa gene encodes a protein kinase that is best known for its role in the regulation of alternative splicing through phosphorylation of multiple SR proteins, among which are TRA, TRA2, and B52 (Nikolakaki et al. 2002). Since B52 is required for splicing of de2f2 pre-mRNA and B52 is a Su(E1) (Rasheva et al. 2006), we considered the possibility that mutation of Doa rescues the de2f1 mutant phenotype through its regulation of B52. So far we have not found evidence to support this model. Inactivation of Doa does not alleviate repression of endogenous dE2F2 targets in eye discs or in tissue culture cells. Additionally, depletion of DOA has no effect on dE2F2-dependent repression in transient transfection assays. Finally, there was no difference in the dE2F2 level in clones of Doa mutant cells or when DOA was depleted by RNAi in S2 cells (data not shown). Thus, it is possible that although DOA phosphorylates B52 in an in vitro kinase assay, B52 may not be an endogenous substrate of DOA. Alternatively, regulation of B52 by DOA may be tissue specific or regulation of B52 by means of phosphorylation in vivo could be directed by multiple SR protein kinases.
In contrast to depletion of L(3)MBT (this work) and B52 (Rasheva et al. 2006), depletion of GFZF and CG31133 does not compromise dE2F2/RBF-mediated repression, but instead we found that expression of the E2F reporter is elevated in these cells. Since the E2F reporter is sensitive to the level of endogenous E2F activity, this raises the possibility that inactivation of GFZF or CG31133 may potentiate dE2F1 activity through, for example, the release of RBF inhibition of dE2F1. Such an explanation seems unlikely since codepletion of RBF together with GFZF or CG31133 exerts an additive effect on the E2F reporter, suggesting that the two Su(E1)'s likely act in parallel to the RBF/E2F pathway. Finally, since both gfzf and CG31133 were isolated in the background of a de2f1 mutation, this disfavors the interpretation that they act through dE2F1. We suggest that CG31133 and GFZF may directly or indirectly cooperate with dE2F1 on a subset of target genes. How precisely GFZF and CG31133 elevate the expression of the E2F reporter is not known and further experiments will be needed to dissect the molecular mechanism of this effect.
Interestingly, the S-phase entry and exit in the SMW are delayed and the S phase appears to be extended in de2f1 gfzf and de2f1 CG31133 double mutant cells. In another example, rescue of the de2f1 mutant phenotype by a mutation in the gene belle is not accompanied by the full restoration of the SMW in de2f1 belle double mutant cells (Ambrus et al. 2007). Similarly, the loss of belle does not relieve dE2F2-mediated repression. In contrast, mutations of l(3)mbt (this work and Lewis et al. 2004; Lu et al. 2007) and B52 (Rasheva et al. 2006) compromise dE2F2-mediated repression while de2f1 l(3)mbt and de2f1 B52 double mutant cells enter and exit the SMW relatively on time. Thus, it is tempting to suggest that relief of dE2F2-mediated repression is needed for de2f1 mutant cells to rescue the correct timing of the SMW. Since prior to the S phase entry in the SMW cells is transiently arrested in G1 within the MF, this may explain why de2f1 mutant cells in the SMW are particularly sensitive to the reduced level of E2F targets. This idea is in agreement with the results of experiments in mammalian cells with a dominant negative form of E2F, which suggested that E2F-dependent activation is needed for cell cycle reentry from quiescence (Rowland and Bernards 2006). Future studies of other isolated suppressors of the de2f1 mutant phenotype are likely to provide novel insights into the growth suppressive function of dE2F2/RBF in vivo.
We are grateful to the Developmental Studies Hybridoma Bank (University of Iowa) and the Bloomington Stock Center for fly stocks and antibodies. We thank D. Knight, K. Marsh, R. Suckling, and H. Summersgill for technical assistance and N. Dyson, A. Katzen, G. Ramsey, and M. Truscott for critical discussions. This work was supported by grant GM079774 from the National Institutes of Health to M.V.F. and by predoctoral fellowship 0815661G from the American Heart Association to A.M.A.
↵1 These authors contributed equally to this work.
Communicating editor: J. Tamkun
- Received May 2, 2009.
- Accepted June 6, 2009.
- Copyright © 2009 by the Genetics Society of America