A Genetic Screen for Dominant Modifiers of a cyclin E Hypomorphic Mutation Identifies Novel Regulators of S-Phase Entry in Drosophila

Abstract

Cyclin E together with its kinase partner Cdk2 is a critical regulator of entry into S phase. To identify novel genes that regulate the G1- to S-phase transition within a whole animal we made use of a hypomorphic cyclin E mutation, DmcycE^JP, which results in a rough eye phenotype. We screened the X and third chromosome deficiencies, tested candidate genes, and carried out a genetic screen of 55,000 EMS or X-ray-mutagenized flies for second or third chromosome mutations that dominantly modified the DmcycE^JP rough eye phenotype. We have focused on the DmcycE^JP suppressors, S(DmcycE^JP), to identify novel negative regulators of S-phase entry. There are 18 suppressor gene groups with more than one allele and several genes that are represented by only a single allele. All S(DmcycE^JP) tested suppress the DmcycE^JP rough eye phenotype by increasing the number of S phases in the postmorphogenetic furrow S-phase band. By testing candidates we have identified several modifier genes from the mutagenic screen as well as from the deficiency screen. DmcycE^JP suppressor genes fall into the classes of: (1) chromatin remodeling or transcription factors; (2) signaling pathways; and (3) cytoskeletal, (4) cell adhesion, and (5) cytoarchitectural tumor suppressors. The cytoarchitectural tumor suppressors include scribble, lethal-2-giant-larvae (lgl), and discs-large (dlg), loss of function of which leads to neoplastic tumors and disruption of apical-basal cell polarity. We further explored the genetic interactions of scribble with S(DmcycE^JP) genes and show that hypomorphic scribble mutants exhibit genetic interactions with lgl, scab (αPS3-integrin—cell adhesion), phyllopod (signaling), dEB1 (microtubule-binding protein—cytoskeletal), and moira (chromatin remodeling). These interactions of the cytoarchitectural suppressor gene, scribble, with cell adhesion, signaling, cytoskeletal, and chromatin remodeling genes, suggest that these genes may act in a common pathway to negatively regulate cyclin E or S-phase entry.

REGULATION of the G1- to S-phase transition by external signals is critical to the decision to proliferate or to differentiate. Progression through G1 phase is controlled by the activity of the Cyclin-dependent ser/thr protein kinases (Cdks) associated with their regulatory Cyclin partners (Ekholm and Reed 2000). In mammalian cells, the G1 cyclins, Cyclin D (D1, D2, and D3) in association with Cdk4(6), and Cyclin E (E1 and E2) in association with Cdk2, play distinct roles in the G1- to S-phase transition. Cyclin D/Cdk4 functions early in G1, while cyclin E/Cdk2 functions at the G1- to S-phase transition, triggering DNA replication initiation and centrosome duplication. In mammalian cells, Cyclin D/Cdk4 and Cyclin E/Cdk2 act to phosphorylate and inactivate the tumor suppressor protein, Retinoblastoma (Rb; Lundberg and Weinberg 1998), which functions by binding to and inactivating the E2F/Dp transcription factor required for the transcription of S-phase genes (Dyson 1998). Binding and phosphorylation of Rb by Cyclin D/Cdk4 and Cyclin E/Cdk2 inactivate Rb, allowing the E2F/Dp transcription factor to function. G1 Cyclin-Cdks are also regulated by the binding of Cdk inhibitory proteins (Sherr and Roberts 1999), such as the p21^CIP1 class of inhibitors, which bind to Cyclin E/Cdk2, inhibiting its activity and leading to G1 arrest.

The key players in the regulation of the G1- to S-phase transition are highly conserved between mammals and Drosophila (Edgar and Lehner 1996). Genetic analysis in Drosophila has shown that both Cyclin E and Cyclin D act to regulate Drosophila Rb (Rbf; Xin et al. 2002). Drosophila Cyclin E is essential for the G1- to S-phase transition during embryogenesis and is downregulated in G1-arrested cells (Richardson et al. 1993, 1995; Knoblich et al. 1994). In contrast, Drosophila Cyclin D primarily acts to regulate cell growth (increase in cell mass) and through the coupling of cell growth to G1- to S-phase progression, stimulates cell proliferation (Datar et al. 2000; Meyer et al. 2000). As in mammalian cells, Drosophila Cyclin E/Cdk2 activity is regulated via a homolog of p21^CIP1, Dacapo, which is required during exit into a terminal G1 arrest prior to differentiation (de Nooij et al. 1996; Lane et al. 1996). Degradation of Cyclin E protein also plays an important role in limiting cell proliferation, and mutations in the ago gene (encoding a homolog of Cdc4, an F-box-containing component of the G1 phase ubiquitin ligase) result in increased Cyclin E protein stability and excessive cell proliferation during eye development (Moberg et al. 2001). However, relatively little is known about the upstream signals that regulate Drosophila cyclin E transcription or the downstream targets of Drosophila Cyclin E/Cdk2 that lead to the initiation of DNA replication within a whole-animal context.

The developing Drosophila eye presents an ideal system to study the relationship between cell proliferation and differentiation. The eye develops from a single cell layer epithelium at the third larval instar stage, where a wave of morphogenesis moves from the posterior to the anterior of the eye imaginal disc (Thomas and Wassarman 1999). Associated with this wave of morphogenesis is the morphogenetic furrow (MF), where the cell cycle becomes coordinated with differentiation. Within and anterior to the MF cells are arrested in G1, while posterior to the MF a subset of cells begins to differentiate into the photoreceptor cell preclusters and the surrounding cells enter a synchronous S phase, after which a subset of these cells undergoes mitosis. Hedgehog signaling has been shown to be important for Cyclin D and Cyclin E expression in this post-MF cell division (Duman-Scheel et al. 2002). Perturbations to the organized arrangement of cell division in the developing eye by, for example, ectopic expression of S-phase inducers, Cyclin E or E2F/Dp, or the negative cell cycle regulators, human p21 or Drosophila Rbf, result in defects in eye development leading to disorganized or rough adult eyes (de Nooij and Hariharan 1995; Richardson et al. 1995; Asano et al. 1996; Du et al. 1996; Xin et al. 2002). The eye phenotypes resulting from overexpression of Cyclin E, E2F/Dp, or Rbf in the posterior differentiating cells of the eye disc have been used as the basis of genetic screens of EMS-mutagenized flies to identify dominant modifiers, revealing novel regulators of the cell cycle (Staehling-Hampton et al. 1999; Boulton et al. 2000; Lane et al. 2000; Duman-Scheel et al. 2002).

A hypomorphic mutation in Drosophila cyclin E, DmcycE^JP, which results in a rough eye phenotype, has provided an opportunity to carry out genetic screens to identify novel genes involved in the regulation of cyclin E expression and function. We have previously shown that DmcycE^JP exhibits a rough eye phenotype due to a reduction in Cyclin E levels and S phases in the developing eye and that this phenotype is sensitive to the dosage of G1- to S-phase genes known to interact with Cyclin E (Secombe et al. 1998). This article reports the results of mutagenesis and deficiency screens to identify genes that dominantly modify the DmcycE^JP rough eye phenotype and presents initial characterization of DmcycE suppressor genes, predicted to act as negative regulators of Cyclin E and/or the G1- to S-phase transition.

MATERIALS AND METHODS

Mutagenesis screen:

For X-ray mutagenesis, 3- to 5-day-old Drosophila males were placed into empty vials (∼100 in each) and treated with 4000 rad of X rays in a CIS Biointernational X-ray machine using a ¹³⁷Cs radiation source (activity 3400 Ci). Mutagenized flies were then allowed to recover for 4 hr with food before being added to 3-day-old virgin females. The flies were turned into new bottles after 2 days and removed after 4 days. EMS mutagenesis was carried out as previously described (Grigliatti 1998). For both EMS and X-ray mutageneses, DmcycE^JP males isogenic on the second and third chromosomes were mutagenized and crossed en masse to b DmcycE^JP females. The progeny from this cross were scored for dominant modification of the DmcycE^JP rough eye phenotype. In addition, F₁ progeny were scored for black-bodied flies to estimate the mutation frequency. From the number of black mutant flies obtained, we calculated that the X-ray mutagenesis frequency was 2.3 × 10⁻³ and the EMS mutagenesis frequency was ∼3 × 10⁻⁴, which are within the ranges described by previous studies (Grigliatti 1998).

Flies selected as having enhanced or suppressed eyes were crossed to a DmcycE^JP strain to ensure that the modification of the DmcycE^JP rough eye phenotype observed initially was heritable and reproducible and then crossed to second or third chromosome balancers to generate stocks. To simplify the screen and stock generation, only interactors that mapped to the second or third chromosome were kept. Once a stock was generated, flies were crossed to a DmcycE^JP strain to ensure that the enhancer or suppressor mutation segregated away from the balancer chromosome. Any mutations that resulted in a dominant eye roughening in the absence of DmcycE^JP were discarded.

For complementation analysis, inter se crosses were carried out between all lethal alleles on each chromosome and allele combinations that resulted in trans-heterozygous lethality to >98% were considered to be within the same gene group.

Genetic mapping of second chromosome genes was carried out using either the b¹, cn¹, bw¹ or al¹, dp^ov1, b¹, pr¹, c¹, px¹, sp¹ multiply marked chromosomes, while third chromosome genes were mapped using the ru¹, h¹, th¹, st¹, cu¹, sr¹, e^s, ca¹ multiply marked chromosomes. The deficiency kit (Bloomington Stock Center) was used for deficiency mapping. Chromosome cytology of third chromosome suppressors was analyzed after Giemsa staining of polytene chromosomes prepared from non-Tubby larvae from a cross of the suppressor (over TM6B) to Canton-S.

DmcycE interactions:

To test X and third chromosome deficiencies for interaction with DmcycE^JP, stocks were generated using balancers that contained the deficiency chromosome and DmcycE^JP and the stock was crossed to homozygous DmcycE^JP flies and progeny containing the deficiency and DmcycE^JP were examined. To examine second chromosome candidate genes for interaction with DmcycE^JP, the candidate gene mutant was recombined onto a marked DmcycE^JP chromosome (using a dp, b, DmcycE^JP, cn, bw chromosome) balanced over CyO and then crossed to a homozygous DmcycE^JP stock and non-Curly flies were examined. X or third chromosome candidate genes were tested for interaction with DmcycE^JP after generating stocks containing the candidate gene mutant and DmcycE^JP, by crossing to b, DmcycE^JP, bw flies and examining non-FM7 or non-TM6B flies. The eyes of at least 50 flies of the appropriate genotype were examined and compared with b, DmcycE^JP, bw/DmcycE^JP flies.

Phenotypic analysis of cyclin E suppressors:

To determine whether a suppressor was acting at the level of S-phase regulation, second chromosome modifiers were crossed to the Curly-Tubby (Cy-Tb) second chromosome balancer, which carries the Tubby dominant larval marker, and were crossed to homozygous DmcycE^JP flies, and non-Tubby larvae were selected for examination of S phases by BrdU labeling. Third chromosome modifiers were balanced over TM6B (marked by Tubby) and crossed to homozygous DmcycE^JP flies and non-Tubby larvae were picked for BrdU labeling. BrdU labeling was carried out as described previously (Secombe et al. 1998). Cyclin E antibody staining was carried out using a polyclonal Cyclin E antibody raised in rats, as previously described (Crack et al. 2002). To determine whether the stage of lethality was before or after the third instar larval stage, each suppressor stock balanced over Cy-Tb or TM6B was examined for the presence of any homozygous modifier (non-Tubby) larvae that survived to or beyond the third larval instar stage. Scanning electron microscopy of adult eyes was carried out as previously described (Secombe et al. 1998).

RESULTS

Identification of X and third chromosome deficiencies that dominantly modify DmcycE^JP:

We have previously demonstrated that the DmcycE^JP rough eye phenotype is sensitive to the gene dose of known cyclin E-interacting genes (Secombe et al. 1998). To obtain an estimate of how many interactors were expected from a random mutagenesis, available X and third chromosome deficiencies were tested to determine how many of these were able to modify the DmcycE^JP phenotype.

A total of 20 suppressor regions and 16 enhancer regions on the X and third chromosomes were identified by generating homozygous DmcycE^JP flies that were also heterozygous for the deficiency chromosome (Table 1). Consistent with results described previously (Secombe et al. 1998), deficiencies removing genes already known to interact with DmcycE^JP such as RBF, roughex, E2F1, and string behaved as expected (Table 1), with the exception of Df(3R)vin2 that removes cyclin A. We have previously shown that cyclin A mutants dominantly enhance DmcycE^JP phenotype (Secombe et al. 1998), while Df(3L)vin2 and the overlapping deficiency Df(3L)vin5 suppressed the DmcycE^JP rough eye phenotype. The most likely explanation for this is that these deficiencies also delete a dose-sensitive suppressor of DmcycE^JP (Table 1). Suppression of DmcycE^JP was also observed with a deficiency (of the region 63F4–64C15) removing the Drosophila cdc4(ago) gene, which encodes an F-box protein of the Skp1-cullin-F-box (SCF) ubiquitin ligase complex involved in Cyclin E protein degradation and can dominantly suppress the DmcycE^JP rough eye phenotype (Moberg et al. 2001).

TABLE 1.

X and third chromosome regions that modify theDmcycE^JP phenotype

Deficiency	Region removed by deficiency	Effect on DmcycEJP	Candidate genes in the region
X chromosome
Suppressors
Df(1)tBA1	1A1 to 2A	Suppression	Rbfa
Df(1)JC19	2F6 to 3C5	Suppression	NC
Df(1)N73 and Df(1)sqh	5D1–2 to 5D5–6	Suppression	roughex, air4 (−)
Df(1)KA14	7F1–2 to 8C6	Mild suppression	lawc (−), air11
Df(1)vN48	9F to 10C3–5	Suppression	dlg
Df(1)C246	11D–E to 12A1–2	Suppression	BAP60b
Df(1)RK4	12F5–6 to 13A9–B10	Suppression	NC
Df(1)sc72b	13F1 to 14B1	Suppression	NC
Df(1)ma13	19A1–2 to 20E–F	Suppression	RpS6
Enhancers
Df(1)4b18	14B8 to 14C1	Enhancement	NC
Df(1)N19	17A1 to 18A2	Enhancement	fused
Third chromosome
Suppressors
Df(3L)HR232 and Df(3L)HR119	63C6 to 63D3	Suppression	sprouty
Df(3L)GN24	63F4–7 to 64C13–15	Suppression	cdc4 (ago)c
Df(3L)vin2 and Df(3L)vin5	68A2–3 to 68D6	Suppression	NC 3.2 region   (See Table 5)
Df(3L)brm11	71F1–4 to 72D1–10	Suppression	Brahma3.5   region (see Table 5)
Df(3L)81K19	73A3 to 74F	Suppression	argos, Abl, Dab, sina,   sinahd3.1 region (Table 5)
Df(3R)p712	84D4–6 to 85B6	Suppression	pyd(Z01)
Df(3R)by10	85D8–12 to 85F1	Suppression	hyd (−)
Df(3R)Cha7	90F1–4 to 91F5	Suppression	NC
Df(3R)XS	96A1–7 to 96A21–25	Mild suppression	NC
Df(3R)T1P	97A to 98A1–2	Mild suppression	l(3)mbt (−), scribble63S15   region (Table 5)
Df(3R)awdKRB	100C6–7 to 100D3–4	Suppression	tramtrack (−)
Enhancers
Df(3L)emc5	61C3–4 to 62A8	Enhancement	emc (−), trio (S) Rac1b
Df(3L)RG7	62B8–9 to 62F2–5	Enhancement	Roughened (Rap1) (−)
Df(3L)HR370	63A1 to 63D10	Enhancement	NC
Df(3L)GN50	63E1–2 to 64B17	Enhancement	cdc2-63E (−), Ras64Bb, RfC40
Df(3L)66CG28	66B8–9 to 66C10	Enhancement	DNApolα50b
Df(3L)hi22	66D10–11 to 66E1–2	Enhancement	h (−), dally (−), mcm7b
Df(3L)AC1	67A2 to 67D7–13	Enhancement	shc, eif-4E, cdk8b
Df(3L)Cat	75B8 to 75F1	Enhancement	Replication-deficient regionb
Df(3L)rdgC	77A1 to 77D1	Enhancement	DNA primase
Df(3L)PcMK	78A3 to 79E1	Enhancement	cyclin H DNApol-etab
Df(3R)Tp110	83C1–2 to 84B1–2	Enhancement	plx
Df(3R)eN19	93B to 94A	Enhancement	E2F1
Df(3R)crbS874	95E8–F1 to 95F15	Enhancement	crb
Df(3R)3450	98E3 to 99A6–8	Enhancement	string

Where deficiencies that overlap have the same effect on DmcycE^JP, the region common to both deficiencies is given as the cytological region. Candidate genes that are underlined have the same effect on the DmcycE^JP phenotype as the corresponding deficiency. Those indicated by (−) have been tested and shown to have no effect on DmcycE^JP, while those indicated by (S) have been shown to suppress rather than enhance. NC, there was no candidate satisfying our criteria in the interval. Gene descriptions are as follows: Rbf, Retinoblastoma; roughex (inhibitor of Cyclin A/Cdk1 in the MF); air4, aberrant immune response 4 (blood cell tumor suppressor); lawc (enhancer of TxG mutants); air11, aberrant immune response 11 (blood cell tumor suppressor); dlg, discs-large (cytoarchetectual protein, neoplastic tumor suppressor); BAP60, Brahma-associated protein 60 (Brahma complex protein, chromatin remodeling); RpS6, Ribosomal protein S6 (translation factor, tumor suppressor); fused (protein kinase required for Hh signaling); sprouty (acts antagonistically to the Egfr); brahma (SWI2-related ATPase, chromatin remodeling, negative growth regulator); argos (Egfr ligand, anatogonist of Egfr signaling); Abl (nonreceptor tyrosine protein kinase); dab, disabled (acts synergistically with Abl); sina and sinah (ring finger E3 ubiquitin ligases, protein degradation); pyd, ZO1, and tamou (membrane-associated guanylate kinase); hyd, hyperplastic discs (HECT domain E3 ubiquitin ligase, protein degradation); l(3)mbt, lethal (3) malignant brain tumor (translation factor, tumor suppressor); scribble (cytoarchitectual protein, neoplastic tumor suppressor); tramtrack (neural differentiation inhibitor); emc, extra-macrochaetae (Id-related HLH repressor protein required for cell proliferation in the wing and with hairy for MF progression in the eye); trio (Rac-GEF, required for Rac activation); Rac1 (Rac family GTPase); Roughened (Rap1; Ras-like GTPase); cdc2-63E (cdc2-related protein kinase); Ras64B (Ras-related); RfC40, Replication factor-C40 (DNA replication initiation); DNApolα50, DNA polymerase-α 50-kD subunit (DNA replication); h, hairy (see emc); dally (glypican, cooperates with Wg and other growth factor receptors); mcm7, minichromosome maintenance 7 (DNA replication initiation); shc (adaptor protein required for Egfr signaling); eif-4E (translational initiation factor); cdk8 (cdc2-related protein kinase); DNA primase (DNA replication); cyclin H (Cyclin required for activation of cdk8 protein kinase); DNApol-eta (DNA replication); plx, pollux (a cell adhesion protein related to the human oncogene TRE17; Zhang et al. 1996); E2F1 (S-phase transcription factor); crb, crumbs (apical-lateral membrane protein involved in cell polarity); and string (Cdc25 phosphatase, activator of Cdc2). An unidentified gene essential for DNA replication is located within the 75B8–75F1 region (Smith et al. 1993). For more details see text.

^aSpecific alleles of RBF have been shown to suppress the DmcycE^JP phenotype and overexpression of RBF enhances the DmcycE^JP rough eye phenotype (Secombe et al. 1998; our unpublished data).

^bNo specific mutation is available to test the interaction.

^ccdc4 (ago) alleles have been shown to suppress DmcycE^JP (Moberg et al. 2001).

^dsina alleles showed only weak dominant suppression of the DmcycE^JP rough eye phenotype; however, a sina-related gene is located next to sina (sinah), and removal of both may account for the suppression observed by the deficiency (M. Coombe, L. Quinn, R. Dickins, J. Secombe and H. Richardson, unpublished results).

For the remaining DmcycE^JP interacting deficiencies, the regions were searched for possible candidate modifying genes using the cytosearch function at Flybase. Candidate cyclin E-interactors were genes expected to either promote S-phase entry for enhancers or inhibit S-phase entry for suppressors. These include homologs of tumor suppressors or oncogenes, genes involved in the initiation of DNA replication, in ubiquitin-mediated degradation pathways, or in chromatin remodeling (Table 1). Genes involved in chromatin remodeling were considered candidates, on the basis of the observation in mammalian cells that components of the SWI/SNF-Brahma chromatin remodeling complex negatively regulate cell proliferation (Harbour and Dean 2001). In a number of cases, specific mutations in these candidate genes were tested for modification of the DmcycE^JP phenotype. This approach enabled the identification of a number of novel cyclin E-interacting genes (Table 1, and see below). For the most part, however, identification of candidates within the modifying deficiency, based on the expected classes of interactors, was not successful. Of the 36 regions that modify DmcycE^JP, candidate genes for only 12 of these were shown to modify DmcycE^JP in a way that would account for the modification by the deficiency. In addition to Rbf1, roughex, ago (cdc4), E2F1, and string discussed above, discs-large (dlg), RpS6, brahma, sina, Abl, scribble, and crumbs were identified in this way and are discussed in detail below. Many interactors did not have an obvious candidate gene within the deficiency breakpoints, or possible candidates were tested but did not interact with cyclin E, or specific mutations were not available in the candidate genes.

Tumor suppressors and oncogenes:

From the DmcycE^JP deficiency screen, a number of regions that showed suppression removed Drosophila tumor suppressor genes, while many that enhanced removed potential oncogenes (Table 1). These candidate genes, as well as other potential oncogenes or tumor suppressors, were specifically tested where possible (Table 2).

TABLE 2.

Interaction of tumor suppressor and signaling pathway mutants withDmcycE^JP

Gene: allele	Function	Effect on DmcycEJP
Tumor suppressors
dlg: dlg6	Cell polarity	Suppression
l(3)mbt: l(3)mbtE2	No effect
brat: l(2)37Cf1	Translation	No effect
air2	No effect
air4	No effect
air6	Mild suppression
air7	Suppression
air8 (RpS6)	Translation	Suppression
air10	Suppression
air13	No effect
hop-air (activated allele of Jak)	Mitogenic signaling	Suppression
air16	Mild suppression
lats/warts: latsP1,91	Protein kinase	No effect
hyd: hyd15	HECT domain Ubiquitin ligase	No effect
fat: fat4,G-rv	Atypical Cadherin	Suppression
expanded: ex01270	FERM domain 4.1 superfamily	Suppression
shotgun: shgk03401,2	E-cadherin	Mild suppression
Negative regulators of signaling pathways
patched: ptcIIIa,G12	Inhibitor of Hh signaling	Suppression
Gap1: Gap11	Ras-GAP—inactivates Ras	Suppressed
yan/aop: aop1	Inhibitor of Ras pathway—Pnt transcription upregulation	No effect
axin: axnE77	Inhibitor of Wg signaling	Mild enhancement
Positive regulators of signaling pathways
Egfr-topQY1—hypomorphic	Receptor tyrosine protein kinase	LOF—enhancement
ElpB1—ligand independent activated allele	(RTK)	GOF—suppression
Ras85D: Ras85e1b	Effector of Egfr signaling	Enhancement
drk: drk10626	Effector of Egfr signaling	Slight enhancement
pointed: pnt7825	Transcription factor—downstream of Ras signaling	Slight suppression
rhoA: rhoA72R,720	Actin cytoskeleton reorganization	Enhancement
Roughened (Rap1): Rap1CD3	Actin cytoskeleton reorganization	LOF—no effect
Notch: N264-39	Signaling protein	No effect
hedgehog: hh18,21,AC	Signaling protein	Mild enhancement
wingless: wgIL1-8,1-17	Signaling protein	Mild suppression
armadillo: armYD35	Transcriptional factor mediator of Wg signaling	Suppression
disheveled: dsh3,6	Mediator of Wg signaling	Mild suppression
spen (poc): poc261-18,361-6	RNA-binding protein	Suppression

LOF, loss of function; GOF, gain of function.

Deficiencies removing potential oncogenes that enhanced DmcycE^JP include those removing a Ras-like GTPase Rac1 (61C3–4; 62A8), a Rap-related GTPase Roughened/Rap1/dRas3 (62B8–9; 62F2–5), and a Ras-like GTPase Ras64B (63E1–2; 64B17). Loss-of-function mutations in Rap1 did not affect DmcycE^JP (not shown) and we have not yet tested Rac1. However, we have shown that trio, which encodes a Rac activator, dominantly suppresses DmcycE^JP (see below). Moveover, Rac2, which plays a redundant role with Rac1, was isolated in a screen for genes that when overexpressed inhibit cell proliferation in the Drosophila eye, which was rescuable by ectopic expression of Cyclin E (Tseng and Hariharan 2002). Taken together these data suggest that Rac is a negative regulator of G1-S progression in Drosophila and thus it is unlikely that halving the dosage of Rac1 accounts for the dominant enhancement of the 61C3–4 to 62A8 region. We were also unable to test Ras64B since there were no available mutants in this gene. However, we tested whether mutants in other oncogenic GTPases, Ras85D and Rho1, could enhance DmcycE^JP (Table 2). Ras has a well-established role in oncogenesis in mammalian cells (Malumbres and Barbacid 2003) and overexpression of an activated form of Ras85D in Drosophila results in a hyperplastic phenotype (Karim and Rubin 1998). Ras85D has also been shown to increase Cyclin E protein levels post-transcriptionally in the wing and eye discs (Prober and Edgar 2000; Brumby and Richardson 2003). Consistent with its expected role as positive regulator of G1-S progression, mutations in Ras85D dominantly enhanced the DmcycE^JP rough eye phenotype (Table 2; data not shown). In mammalian cells, Rho promotes cell proliferation and is required for Ras-induced transformation (Sahai and Marshall 2002). Indeed, overexpression of wild-type and dominant active forms of mammalian Rho have been shown to upregulate Cyclin E/Cdk2 activity and induce progression from G1 into S phase. Although, no role for Rho1 has been revealed in G1-S progression in Drosophila, we observed that mutants in rho1 dominantly enhanced the DmcycE^JP rough eye phenotype (Table 2; data not shown), revealing a novel role for Drosophila Rho1 that warrants further investigation.

In addition, a deficiency removing fused (17A1–18A2), an effector of the Hh pathway, showed enhancement of DmcycE^JP. Although we have not specifically tested fused to determine whether it represents the interacting gene, this interaction is consistent with the recent observation that the Hedgehog (Hh) pathway acts to upregulate cyclin E transcription in the eye (Duman-Scheel et al. 2002) and that upregulation of the Hh pathway is oncogenic in mammals (Wetmore 2003). To explore this further, we analyzed the effect of halving the dose of Hh or patched (a negative regulator of the Hh receptor, Smoothened) on the DmcycE^JP rough eye phenotype (Table 2; data not shown). As expected, Hh alleles dominantly enhanced while patched alleles dominantly suppressed DmcycE^JP, consistent with a role for the Hh pathway in positively regulating cyclin E and inducing S-phase entry. We also examined other signaling pathways for dominant interactions with DmcycE^JP (Table 2; data not shown). In mammalian cells, the EGF receptor, the Wnt/Wingless, and Notch signaling pathways have a growth and/or cell cycle stimulatory role in many cells and can be oncogenic when upregulated (Allenspach et al. 2002; Chang et al. 2003; Giles et al. 2003). Consistent with the interaction of Ras85D with DmcycE^JP, loss-of-function mutations in the EGF receptor (Egfr) enhanced DmcycE^JP while gain-of-function mutations (Ellipse) suppressed. Other downstream components of the Egfr-Ras pathway also interacted with DmcycE^JP in a manner consistent with the Egfr having a positive role in regulating Cyclin E and entry into S phase (Table 2). Reducing the dose of Notch, however, showed no effect on the DmcycE^JP phenotype. Interestingly, halving the dosage of wingless (wg), disheveled (encoding a Wg-signaling mediator), and armadillo (arm; encoding a β-catenin homolog, the Wg signaling transcriptional effector) resulted in suppression of DmcycE^JP. In contrast, halving the dosage of axin (encoding an inhibitor of Wg signaling) enhanced DmcycE^JP. While contrary to the expected role of the Wg pathway, an inhibitory proliferative function for Wg has been observed in the zone of nonproliferation in the third instar wing pouch (Johnston and Edgar 1998; Johnston et al. 1999; Johnston and Sanders 2003). Similarly, we have previously shown that the Dpp (TGFβ homolog), although growth stimulatory earlier in development, acts to negatively regulate cell cycle progression in the third instar eye imaginal disc and mutants that disable the Dpp signaling pathway dominantly suppress DmcycE^JP (Horsfield et al. 1998).

Several DmcycE^JP suppressor regions on the X chromosome and on the third chromosome remove known Drosophila tumor suppressor genes. Specific mutations were available for some of the candidate genes encoding tumor suppressors and were therefore tested for a genetic interaction with cyclin E. Specific mutations in Ribosomal protein S6 (RpS6 air8), the best candidate for the cyclin E suppressor in the 19A–20F region, were tested and shown to suppress the DmcycE^JP rough eye phenotype (Table 2; not shown). Mutations in RpS6 were identified as loss-of-function mutations that result in overproliferation of larval hematopoietic tissues and give rise to variable melanotic tumor phenotypes (Gateff et al. 1996). RpS6 is phosphorylated in response to mitogen stimulation and phosphorylated RpS6 is preferentially incorporated into polysomes, resulting in an increased rate of translation of a subset of transcripts (Amaldi and Pierandrei-Amaldi 1997; Martin and Blenis 2002). However, disruption of Drosophila S6 kinase leads to reduced growth and smaller flies and mutation of the upstream kinase Tor causes cell cycle arrest that can be rescued by cyclin E expression (Zhang et al. 2000). Furthermore, conditional knockout of RpS6 in mice results in a specific block in cyclin E expression (Volarevic et al. 2000). Given this role for RpS6 in mammalian cells, it is unknown how halving the dosage of RpS6 leads to the suppression of DmcycE^JP; however, it is consistent with the tumor suppressor function of Drosophila RpS6.

Other Drosophila tumor suppressors were tested for interaction with DmcycE^JP (Table 2), and those that showed suppression were hop-air (an activating mutation in JAK kinase), consistent with a role for Drosophila Jak in cell proliferation and that Cyclin D-Cdk4 and Cyclin E-Cdk2 bind and regulate STAT92E protein stability (Chen et al. 2003); fat (encoding an atypical Cadherin involved in planar polarity); expanded (encoding a FERM domain protein involved in actin remodeling); and the unidentified air7, air10, and air16 (Gateff et al. 1996; de Lorenzo et al. 1999). The Drosophila E-cadherin gene, shotgun (shg; Tepass et al. 1996; Uemura et al. 1996), when halved in dosage, was also shown to slightly suppress DmcycE^JP. In contrast, lethal (3) malignant brain tumor [l(3)mbt] and hyperplastic discs (hyd; Gateff et al. 1996; de Lorenzo et al. 1999), which were considered candidates for the regions 97A–98A2 and 85D8–85E13, respectively (Table 1), did not modify the DmcycE^JP phenotype when specific alleles were tested (Table 2; data not shown). Taken together these data suggest that there are specific pathways that show rate-limiting effects on Cyclin E and thereby entry into S phase, in the eye imaginal disc.

Identification of cyclin E interactors using a mutagenic dominant modifier screen:

As described above, screening for dominant genetic modifiers of DmcycE^JP using deficiencies and candidate gene approaches has revealed some interesting interactors. However, this approach is limited in that the deficiencies may remove more than one modifier, confounding the identification of interacting genes. For these reasons, an unbiased genetic screen for DmcycE^JP modifiers using mutagenized flies was carried out, to generate specific modifier mutations that could be further characterized. To randomly generate mutations that could then be examined for their effect on the DmcycE^JP phenotype, we utilized X-ray mutagenesis, which causes deletions and chromosomal rearrangements (Sankaranarayanan and Sobels 1976) that are expected to aid in the identification of the modifier, and EMS mutagenesis, which causes nucleotide substitutions resulting in missense or nonsense mutations (Lifschytz and Falk 1968). For the X-ray mutagenesis, 39,234 F₁ flies were screened for modification of the DmcycE^JP rough eye phenotype and stocks of 104 suppressors and 59 enhancers that consistently modified the DmcycE^JP phenotype on the second or third chromosomes were generated (summarized in Table 3). For the EMS mutagenesis a total of 15,049 F₁ flies were screened and 29 suppressors and 54 enhancer mutations on the second or third chromosomes were isolated (Table 3).

TABLE 3.

Summary of the 246 modifiers identified in the screen

	Second chromosome		Third chromosome
	Homozygous lethal	Homozygous viable	Homozygous lethal	Homozygous viable
Suppressors
EMS	19	5	5	0
X ray	60	12	26	3
Enhancers
EMS	13	1	38	2
X ray	13	8	29	9

Summary of the number of homozygous viable and homozygous lethal second and third chromosome modifiers obtained from the EMS and X-ray mutageneses. Not included are three X-ray-generated suppressor mutations likely to be translocations to the Y, for which it was not possible to know whether they were homozygous viable or lethal.

DmcycE^JP suppressor complementation groups:

For the second chromosome homozygous lethal suppressors, complementation analysis revealed that there were 10 complementation groups containing more than one allele, as well as many with single alleles (Tables 4 and 5; and data not shown). In addition, these stocks were crossed to a number of alleles on the second chromosome identified in the screens for enhancers of the eye phenotypes generated by overexpression of cyclin E or E2F1/Dp (Staehling-Hampton et al. 1999; Lane et al. 2000). This analysis revealed that 62S9 was allelic to E(sev-cycE)^e93 (and was termed group 2.11). Further analysis revealed that some members of group 2.6 contained a second lethal allele that was distinct from the lethal common to group 2.6 members, forming two new groups, 2.12 (containing the 2.6 allele, 42S13, and a single allele 22S9) and 2.13 (containing the 2.6 alleles 42S14 and 66S4 and the 2.7 allele 55S2). Thus there were a total of 13 second chromosome suppressor groups with multiple members. For the third chromosome suppressors, complementation analysis revealed that there were 5 groups containing >1 allele, and there were many single alleles (Tables 4 and 5). Groups 3.3 and 3.4, however, cannot truly be considered as groups with more than one allele as there were only two members in each and they both contained a common member, 65S55, which appears to contain a large deletion. The suppression of the DmcycE^JP adult eye phenotype by representatives of the identified suppressor groups is shown in Figure 1.

TABLE 4.

Summary of identified suppressors

Group	Alleles	Cytological location	Stage lethal	Identified genes
Second chromosome
2.1	23S9, 27S3, E2S31, E6S2	Genetically linked to aristaless (21C)	Third larval instar	lethal-2-giant larvae (lgl)
FTC: Df(2L)net-PMF
C: Df(2L)al
21A1–B8
2.2	28S2, 38S4, 39S2	Genetically between cinnabar (43E)   and brown (59E)	Embryonic	phyllopod
FTC: Df(2R)trix
C: Df(2R)3072r and Df(2R)CX1
51A1–5
2.5	42S11, 58S12	FTC: Df(2R)nap9 and Df(2R)ST1	Pupal	l(2)04524 dEB1
C: Df(2R)nap1 and In(2R)pk78s
42B3–42C7
2.11	62S9, E(sev-cycE)e93	FTC: Df(2R)Jp1 and Df(2R)XTE18	Embryonic	l(2)01288 scab   (α-integrin)
C: Df(2R)03072 and Df(2R)Jp4
51D3–51F13
2.12	42S13, 22S9	FTC: Df(2L)TW137, Df(2L)H20,   and Df(2L)M36F-55	Larval/pupal	cadNa
C: Df(2L)TW50 and Df(2L)TW3
36D1–E4
Third chromosome
3.5	25S14, E6S8	FTC: Df(3R)brm11b	Embryo	brahma
C: Df(3R)BK10 and Df(3R)st-f13
71F1–72D1
2S1	FTC: Df(3L)emc5c	Before third larval   instar	trio (Rac GEF)
C: Df(3L)emc-E12, Df(3L)Ar11, and Df(3L)RG5
61E–62A8
35S1	Genetically between stripe (86D1) and   curled (90F7)	Embryo	moira
FTC: Df(3R)sbd-105d
C: Df(3R)ea and Df(3R)PO4
89A11–89B10
43S2	Genetically left of scarlet (73A4)	Pupal	l(3)72Dk zn72D
FTC: Df(3L)st-f13,d Df(3L)st-g24,d   and Df(3L)th102d
C: Df(3L)st-b11 and Df(3L)brm11
72D1–72D10
63S15	FTC: Df(3R)T1P,b Df(3R)Tl-X, and Df(3R)Tl-I	Third larval instar	scribble
C: Df(3R)XTAI and Df(3R)3450
97B–97D2
65S19	Genetically between veinlet (62A) and   thread (72D)	Semilethal	trithorax-like
FTC: Df(3L)fz-M21d
C: Df(3L)fz-GR3b and Df(3L)BK10
		70D4–71C3

Underlined alleles are members of more than one group. FTC, failed to complement; C, complemented.

^aNot confirmed by testing specific allele for suppression of DmcycE^JP.

^bDeficiency also suppressed DmcycE^JP (Table 1 and data not shown).

^cDeficiency did not suppress and in fact enhanced DmcycE^JP (see Table 1).

^dDeficiency did not suppress DmcycE^JP (data not shown).

TABLE 5.

Summary of unidentified suppressors

Group	Alleles	Cytological location	Effect on DmcycEJP eye disc S phases	Stage lethal
Second chromosome
2.3	59S16, 65S12	Genetically between black (34D) and   cinnabar (43E)	Increased	Pupal
FTC: Df(2L)TW137, Df(2L)TW50, Df(2L)E71,  Df(2L)TW3 and Df(2L)OD15
C: Df(2L)H20 and Df(2L)PR-A16
36F7–37B8
2.4	26S8, 57S6, 59S3	Genetically between black (34D) and   cinnabar (43E)	Increased	Postembryonic
FTC: In(2R)bwVDe2L and Df(2R)nap1	Before third larval   instar
C: Df(2R)nap9
41D2–42A2
2.6	41S1, 42S7, 42S13,   42S14, 65S4, 66S4,   67S7, E3S17, E3S18,   E3S31	Unknown	Increased	Larval/pupal
2.7	55S2, 64S19, 65S39   (14S3, 57S1, 65S23,   E10S15)	FTC: Df(2R)M60E	Increased	Before third instar   larvae
C: Df(2R)ES1
60E2–8
2.8	E6S4,a E6S19	FTC: Df(2L)Dwee1Δ5 and Df(2L)spdj2	Increased	Larval lethal
C: Df(2L)Dwee1w05, Df(2L)J-H, and Df(2L)E110
27B2–C3
2.9	25S11, E1S4	FTC: Df(2R)X58-12 and Df(2R)X58-8	Increased	Postembryonic
C: Df(2R)X58-7, Df(2R)59AD, and   Df(2R)pu-D17	Before third larval   instar
58E4–59A
2.10	65S5, 65S13, E10S34b	FTC: In(2R)bwVDe2lCyR	Increased	Embryonic
C: Df(2R)M41A4 and Df(2R)nap9
41A–41E1
2.13	42S14, 66S4, 55S2	FTC: Df(2R)M60E	Increased	Larval/pupal
C: Df(2R)ES1
60E2–8
Third chromosome
3.1	19S5, 24SX, 58S5, 62S2	Genetically left of thread (72D)	Increased	Larval
FTC: Df(3L)81k19d
C: Df(3L)st-b11 and Df(3L)W10
73D–74F
(male recombination mapping to 74B1–B4)
3.2	41S13, 44S18, 59S26,   65S1, E6S25, E8S13	FTC: Df(3L)lxd6, Df(3L)vin5,d   and Df(3L)vin2d	Increased	Pupal
C: Df(3L)vin7 and Df(3L)AC1
68A9–68B3
3.3	34S3, 65S55	Genetically between veinlet (62A) and hairy   (66D10)	Increased	Larval
62A–66D10e
3.4c	1S2	Genetically left of scarlet (73A3)	Increased	Before third instar   larval
65S55	FTC: Df(3L)29A6f and Df(3L)Rd1-2d
C: Df(3L)AC1
66F5
Breakpoint at 66F
Third chromosome, single alleles
1S3	Genetically right of ebony (93D2)	Increased	Before third instar
FTC: Tp(3;Y)J55	larval/pupal
C: Df(3R)T1-P, Df(3R)D605, Df(3R)3459,   Df(3R)Dr-rv1, Df(3R)L127, and Df(3R)B81
98A–100B
(98A5–98E3)
Breakpoint at 98C
13S1	Genetically right of curled (86D1)	ND	Before third instar
FTC: Df(3R)Dl-BX12f	larval
C: Df(3R)Cha7, Df(3R)KX18, and Df(3R)FX3
91F5–91F11
20S1	Genetically to the left of hairy   (66D10)	Increased	Before third instar   larval
61A–66D10e
Breakpoint at 63E
42S12	Genetically to the left of hairy   66D10)	Increased	Before third instar   larval
61A–66D10e
Breakpoint at 62E/F
42S33	Genetically to the left of thread (72D)	Increased	Second instar larval
61A–72De
43S1	Unknown	Increased	Before third instar   larval
47S8	Unknown	Increased	Before third instar   larval
59S9	Genetically between veinlet (62A) and   thread (72D)	Increased	Semilethal
FTC: Df(3L)RG7g
C: Df(3L)Aprt-1 and Df(3L)M21
62D2–62F5
Breakpoint at 62B
59S18	FTC: Df(3L)RG5f	Increased	Before third instar   larval
C: Df(3L)Aprt-1, Df(3L)Aprt-32, and Df(3L)RG7
62A10–62B1
68S10	FTC: Df(3R)3-4f	Increased	Before third instar
C: Df(3R)110 and Df(3R)e1025-14
82F3–82F10
E9S1	Genetically between veinlet (62A) and   thread (72D)	Increased	Pupal
		62A–72De

Underlined alleles are members of more than one group. 55S2 is a member of 2.7 and 2.13, but other 2.7 alleles complement the 2.13 alleles, 42S14 and 66S4. Weak alleles are in parentheses. These gave escapers that showed the Rehow phenotype with other 2.7 alleles, although they failed to complement Df(2R)M60E. FTC, failed to complement; C, complemented; ND, not determined.

^aE6S4 also contains another lethal at 29D1–2 to 30C4–D1.

^bAll three alleles failed to complement each other, but 65S13 and E10S34 were not completely lethal over the deficiency and gave rise to escapers with rough eyes and wing defects.

^cMapping data are for 1S2.

^dDeficiency also suppressed DmcycE^JP (Table 1 and data not shown).

^eThe given cytological interval was determined only by genetic mapping of the lethal. No deficiencies uncovering this mutant were identified by deficiency mapping.

^fDeficiency did not suppress DmcycE^JP (data not shown).

^gDeficiency did not suppress but rather enhanced DmcycE^JP (see Table 1).

Figure 1.—

The identified dominant suppressors of DmcycE^JP: scanning electron-micrographs of adult eyes and BrdU labeling of eye imaginal discs from DmcycE^JP individuals heterozygous for the identified suppressor alleles. Genotypes are as indicated: wild type (WT); DmcycE^JP; DmcycE^JP; 43S2/+; DmcycE^JP; zn72D/+; DmcycE^JP, 2.2-39S2/+; DmcycE^JP, phyl²²⁴⁵/+; DmcycE^JP; 2S1/+; DmcycE^JP; trio^M89/+; DmcycE^JP, 2.5-42S11/+; DmcycE^JP, dEB1-l(2)04524/+; DmcycE^JP, 2.11-62S9/+; and DmcycE^JP, 2.11-l(2)01288/+.

Complementation crosses revealed genetic interactions between many DmcycE^JP modifiers. Some alleles when trans-heterozygous showed reduced numbers and/or a striking phenotype characterized by rough eyes, held-out wings, and poor viability [the rough eyes and held-out wings (Rehow) phenotype] and in some cases the trans-heterozygous females were sterile. The Rehow phenotype occurred between the severe group 2.7 alleles (55S2, 64S19, and 65S39) and the weak 2.7 alleles (65S23, E10S15) or the single alleles 19S3, 40S5, 42S3, 64S10, 61S10, or 62S9 (group 2.11). These single alleles also showed the Rehow phenotype when crossed with each other. The trans-heterozygous Rehow phenotype of 61S10 and the severe 2.7 alleles was dependent on the presence of the DmcycE^JP mutant, since it was observed only when DmcycE^JP was homozygous. For 55S2, 64S19, 65S39, 19S3, 40S5, 42S3, 62S9, 65S23, E10S15, or 64S10, the Rehow phenotype occurred independent of DmcycE^JP homozygosity (in the background of DmcycE^JP/+). Crosses between the third chromosome single-allele suppressors; 47S8 and 59S9, 20S1, 63S15, or 65S19; also gave rise to the Rehow phenotype. In these cases, genetic and deficiency mapping data suggest that none of these alleles are weak alleles of the same complementation group.

Basic characterization was then carried out on second chromosome multimember complementation groups and most of the third chromosome suppressors (summarized in Tables 4 and 5). We determined whether the suppression of DmcycE^JP was occurring at the level of S phases during eye development (shown for representatives of the identified suppressors in Figure 1). In all cases examined, there was a significant increase in the size of the eye disc as well as in the number of S phases in the anterior and the post-MF S-phase band. Thus all suppressors tested act to suppress DmcycE^JP by increasing S phases in the normal pattern. The stage of lethality of the homozygous modifier mutation was also determined by counting the number of hatched embryos and examining whether homozygous third instar larvae were present (Tables 4 and 5). This analysis led to the observation that group 2.1 and 63S15 homozygotes died as overgrown larvae, a phenotype that occurs with Drosophila neoplastic tumor suppressors (Gateff et al. 1996; de Lorenzo et al. 1999; see below).

Mapping and identification of DmcycE^JP suppressors:

The cytological location of the lethal mutation for the complementation groups and some of the single alleles was determined by crossing suppressors to the deficiency collection (Bloomington Stock Center). In addition, a crude map position was determined for most of the third chromosome interactors and some of the second chromosome interactors by genetic mapping of the DmcycE^JP suppressor mutation. In all cases tested, the map location of the suppressor by genetic mapping was consistent with the map location of the lethal by deficiency mapping. In some cases, chromosome cytology was examined to map aberrations (Tables 4 and 5). Knowledge of the location of the modifier gene then enabled likely candidate genes to be investigated by testing mutant alleles, where available, for failure to complement the modifier mutant.

This strategy enabled the identification of 5 of the 13 second chromosome (2.1, 2.2, 2.5, 2.11, and 2.12) and 6 of the 20 third chromosome suppressors (3.5, 2S1, 35S1, 43S2, 63S15, and 65S19; summarized in Table 4). Of the remaining groups, although map positions were well defined for 7 of 13 of the second (2.3, 2.4, 2.7, 2.8, 2.9, 2.10, 2.13) and 8 of 20 third chromosome suppressor groups (3.1, 3.2, 3.4, 1S3, 13S1, 59S9, 59S18, 68S10), and available candidate gene alleles were tested for each of the suppressor genes, the identity of the suppressors is not yet known (Tables 5 and 7). In these cases, it is likely that these suppressor mutations define novel genes. For two of the suppressor groups (2.6 and 3.3) and five of the single alleles (20S1, 42S12, 42S33, 43S1, 47S8) a precise location for the suppressor was not determined, since none of the available deficiencies failed to complement the suppressor. In these cases the lethal mutation must map to a region not covered by the deficiency collection. A brief description of the identification of the more precisely localized suppressors is detailed below and summarized in Tables 4 and 5.

The identified suppressors:

By complementation tests to known gene alleles the identities of five second chromosome suppressors (2.1, 2.2, 2.5, 2.11, 2.12) and six third chromosome suppressors (3.5, 2S1, 35S1, 43S2, 63S15, 65S19) were revealed. These suppressor genes fall into the functional groups of chromatin remodeling and transcription factors (four genes), signaling (two genes), cytoskeletal (one gene), cell adhesion (two genes), and neoplastic tumor suppressors (two genes). Specific details on the verification and characterization of these suppressors are discussed under these functional groupings (summarized in Table 4).

Chromatin remodeling and transcription factor genes:

3.5. (Brahma):

3.5 was mapped to 71F1–72D1 (Table 4) and alleles of brahma, a SWI2 homolog, encoding a component of the SWI/SNF chromatin-remodeling complex (Papoulas et al. 1998), failed to complement both 3.5 alleles. Furthermore, Df(3R)brm11 (71F1–4; 72D1–10) was identified as a dominant suppressor of DmcycE^JP in the screen of third chromosome deficiencies (Table 1). Consistent with suppressor 3.5 being brahma, we showed that previously isolated alleles of brahma also dominantly suppressed DmcycE^JP (Brumby et al. 2002).

35S1 (Moira):

35S1 was mapped to 89A11–89B10 (Table 4). Candidate mutants in this region were tested for allelism with 35S1, and alleles in moira, a SWI3 (BAP155) homolog, failed to complement 35S1. Consistent with the suppressor 35S1 being moira, we demonstrated that previously isolated alleles of moira also dominantly suppressed DmcycE^JP (Brumby et al. 2002).

Brahma and Moira are components of the Drosophila Brahma (SWI/SNF-related) chromatin remodeling complex (Papoulas et al. 1998), which has been shown to play a role in negatively regulating S phase (Staehling-Hampton et al. 1999; Harbour and Dean 2000). Consistent with this notion, alleles of other Brahma complex genes, snr1 and osa, as well as a deficiency that removes the brahma-associated protein 60 (BAP60) or BAP111, dominantly suppress the DmcycE^JP phenotype (Brumby et al. 2002; Table 1).

65S19 (Trithorax-like):

65S19 was only semilethal; however, genetic and deficiency mapping was still possible, and 65S19 was located to 70D4–71C3 (Table 4). Complementation tests of candidate genes in the region revealed that Trithorax-like (Trl) was allelic to 65S19. Consistent with this, previously characterized alleles of Trl also dominantly suppressed DmcycE^JP (Brumby et al. 2002).

43S2[l(3)72Dk (zn72D)]:

43S2 was localized to 72D1–72D10 (Table 4) and complementation tests of mutations in the 72D1–10 region revealed that In(3)Taf4^XS-2884, an inversion affecting expression of Taf4 (Taf110) and Zn72D (Sauer et al. 1996), failed to complement 43S2. A specific EMS allele of Taf4, l(3)72Dj, however, complemented 43S2, suggesting that 43S2 is most likely allelic to zn72D (CG5215). Indeed, another EMS allele in the region, l(3)72Dk, which failed to complement In(3)Taf4^XS-2884, also failed to complement 43S2, suggesting that l(3)72Dk is an allele of zn72D. The zn72D gene encodes a zinc finger protein, but has not been characterized. In an attempt to verify the identity of 43S2 suppression as being due to a mutation of zn72D, l(3)72Dk was crossed into the DmcycE^JP background. However, l(3)72Dk did not suppress the DmcycE^JP adult eye phenotype or the S-phase defect of DmcycE^JP eye discs as effectively as 43S2 did (Figure 1), which may be due to l(3)72Dk being a weaker allele than 43S2. Molecular characterization of the 43S2 and l(3)72Dk lesion will be required to confirm this. Interestingly, Zn72D was identified in a differential expression screen as a gene expressed specifically in the differentiating region of the eye disc (Jasper et al. 2002), consistent with a role for Zn72D in cell cycle arrest or differentiation.

Signaling pathway genes:

2.2. (phyllopod):

2.2 was localized to 51A1–51A5 (Table 4). Consistent with this, Df(2R)trix (51A1–2; 51B6) dominantly suppressed the DmcycE^JP rough eye phenotype (data not shown). Mutations and P alleles within the 51A region were tested for allelism with 2.2 alleles, revealing that a null allele of phyllopod, phyl²²⁴⁵, failed to complement all three S(DmcycE^JP) 2.2 alleles. To verify that 2.2 was indeed phyl, previously identified phyl alleles (2245 and 2366) were tested and shown to dominantly suppress the rough eye phenotype and the S-phase defects of DmcycE^JP (Figure 1; M. Coombe, L. Quinn, R. Dickins, J. Secombe and H. Richardson, unpublished results). These data are consistent with the mutation of phyl being responsible for the observed suppression of DmcycE^JP by the 2.2 alleles.

Phyl expression is induced by the Sevenless receptor tyrosine kinase signaling pathway and is a rate-limiting component in R7 photoreceptor cell differentiation in the eye imaginal disc, but also has other roles in neural differentiation during development (Dickson 1998). Phyl is a pioneer protein (containing no homology to other known proteins) that functions with the Ring finger protein Seven in absentia (Sina) and the F-box protein Ebi, to bind to and target the two isoforms of the neural differentiation inhibitor, Tramtrack (Ttk69 and Ttk88) and probably other proteins for destruction by the ubiquitin/proteosome pathway, allowing neural cell differentiation (Li et al. 1997; Tang et al. 1997; Boulton et al. 2000). Consistent with this, homozygous viable mutants in sina (sina¹) strongly suppressed the DmcycE^JP adult rough eye and S-phase defects, while a stronger sina allele (sina²) showed weak dominant suppression (M. Coombe, L. Quinn, R. Dickins, J. Secombe and H. Richardson, unpublished results). However, a deficiency removing sina showed strong dominant suppression of DmcycE^JP (Table 1). This deficiency removes a sina-related gene (sina-h), located adjacent to sina, as well as Abl, which has been shown to dominantly suppress DmcycE^JP (see below). Consistent with the involvement of the Sina complex in negative regulation of G1-S, ebi alleles have been shown to dominantly suppress DmcycE^JP (Boulton et al. 2000). The mechanism by which the Sina complex acts to regulate G1-S does not involve targeting Cyclin E or E2F for ubiquitin-dependent degradation (Boulton et al. 2000) and remains to be determined.

2S1 (trio):

2S1 was mapped to 61E–62A8 (Table 4) and by crosses to mutations within the region it was revealed that trio [encoding a Rac guanine nucleotide exchange factor (Rac-GEF; Bateman et al. 2000)] failed to complement 2S1. To confirm this interaction, a previously isolated allele of trio (trio^M89) was crossed into the DmcycE^JP background. trio^M89 was shown to dominantly suppress the DmcycE^JP rough eye phenotype and S-phase defect (Figure 1). Rac-GEFs are involved in the activation of Rac family GTPases, which have roles in actin cytoskeletal remodeling (Blanchard 2000). In mammalian cells, Rac can lead to repression of Rho activity (Sander et al. 1999), and therefore mutation of trio may lead to higher levels of Rho activity. Rho activation in mammalian cells has been shown to promote cell cycle progression by leading to downregulation of the Cyclin/Cdk inhibitors p21 and p27 (Aznar and Lacal 2001; Pruitt and Der 2001; Sahai and Marshall 2002). trio has been shown to genetically interact with Abl, encoding a nonreceptor tyrosine kinase also involved in actin cytoskeleton remodeling (Luo 2000). Consistent with this, the deficiency removing Abl (73A3; 74F) dominantly suppressed the DmcycE^JP rough eye phenotype; however, this deficiency also removes sina, sina-h (see above), and the Abl pathway gene, Disabled (Dab). The Abl alleles Abl⁰⁴⁶⁷⁴ and Abl¹ were then tested and shown to also suppress the DmcycE^JP rough eye phenotype (not shown). The precise mechanism by which reducing the dosage of trio and Abl leads DmcycE^JP suppression remains to be determined.

Cytoskeletal genes:

2.5. (dEB1):

2.5 was localized to 42B3–42C7 (Table 4). 2.5^58S12 was also lethal over the adjacent deficiency, Df(2R)nap1 (41D2–E1; 42B1–3), indicating that this allele is a deficiency or rearrangement that affects a larger region than 2.5^42S11. S(DmcycE^JP)2.5^42S11 was crossed to P-element alleles available in the region and l(2)04524, was found to be semilethal in combination with 2.5^42S11. The few escaper flies, trans-heterozygous for 2.5^42S11 and l(2)04524, did not have any gross abnormalities, but generally died within a few days of eclosing, and the females were sterile. l(2)04524 is inserted within the 5′-UTR of the Drosophila homolog of the EB1 gene (BDGP). dEB1 encodes a cytoskeleta1 protein that binds to microtubules and plays an important role in adherens junction integrity and cell polarity (Lu et al. 2001; Rogers et al. 2002). EB1 was identified in mammalian cells as a binding partner of the adenomatous polyposis coli (APC) colon cancer tumor suppressor (Su et al. 1995); however, Drosophila APC1 and APC2 both lack the EB1-binding domain. Consistent with the identity of 2.5 being dEB1, l(2)04524 and the EMS dEB1 alleles, dEB1⁵ (1DL) and dEB1⁶ (GJ63/9) (obtained from J. Roote), dominantly suppressed DmcycE^JP rough eye and S-phase defects (Figure 1 and data not shown). Moreover 2.5^42S11 and l(2)04524 disrupt dEB1 transcription (D. Coates, L. Quinn, R. Dickins, J. Secombe, A. Brumby and H. Richardson, unpublished results). How the EB1 microtubule protein is involved in G1-S regulation remains to be determined.

Cell adhesion genes:

2.11. (scab) (α-Integrin):

Group 2.11 was defined by S(DmcycE^JP)62S9 from this screen and E(sev-cycE)^e93 was from the Lane et al. (2000) genetic screen (see above). 2.11 was mapped to the region 51D3–51F13 (Table 4), and by testing mutations within this region, it was revealed that the P allele, l(2)01288, failed to complement both 2.11 alleles. The insertion point of l(2)01288 has been defined (BDGP) and disrupts the scab gene, encoding an α-integrin, αPS3, thought to play a role in tissue morphogenesis (Stark et al. 1997). To further confirm that 2.11 is allelic to scab, previously identified EMS-derived alleles of scab (scb¹ and scb²) were tested and shown to also fail to complement 2.11 alleles. Consistent with the suppressing gene being scab, l(2)01288, scb¹, and scb² were recombined onto the DmcycE^JP and were shown to also suppress the rough eye phenotype and the S-phase defect of DmcycE^JP (Figure 1 and data not shown). In mammalian cells, integrins in association with the extracellular matrix have a well-established role in promoting anchorage-dependent cell proliferation (Danen and Yamada 2001). However, recent studies have shown that integrins can also inhibit G1-S progression (Hazlehurst et al. 2000; Mettouchi et al. 2001). Our identification of scab in the DmcycE^JP genetic screen suggests that in Drosophila integrins also act as negative regulators of G1-S.

2.12. (CadN):

2.12 alleles 42S13 (also an allele of group 2-6) and 22S9 (Figure 1 and data not shown) were mapped to 36D1–36E4 (Table 4). Mutations and P alleles in the region were tested by complementation analysis, revealing that an allele of CadN (CadN^M12) failed to complement both 2.12 alleles. CadN encodes a cadherin-like transmembrane protein (Lee et al. 2001; Iwai et al. 2002) that can bind to α-catenin and β-catenin (Armadillo), components of the adherens junction (Perez-Moreno et al. 2003). In mammalian cells, downregulation of N-Cadherin leads to upregulation of G1 Cyclin activity (Charrasse et al. 2002). Due to the close location of CadN and DmcycE, it was not possible to obtain a recombinant of the CadN allele with DmcycE^JP to confirm that CadN exhibits the same modifier effect as S(DmcycE^JP)2.12.

Cytoarchitectural tumor suppressor genes:

2.1. [lethal-(2)-giant larvae]:

2.1 was localized to 21A1–21B7–8 by deficiency mapping (Table 4). The mapping of 2.1 was initially confounded by the fact that two deficiencies in the deficiency kit, Df(2L)Prl (32F1–3; 33F1–2) and Df(2L)J39 (31D1–11; 32D1–E5), also contained lesions in the 21A region and therefore failed to complement 2.1. The localization of 2.1 was confirmed by genetic mapping of 2.1 alleles, which indicated that the lethal mapped to the left of UbcD1 (32A4–5) and close to al (21C2–4). Since 2.1 homozygous mutants die as giant larvae, an allele of the lethal-(2)-giant-larvae (lgl) gene, which also gives giant larvae and is localized at 21A, was tested for complementation of 2.1 alleles and failed to complement, whereas mutations in other genes in this region that have been identified as negative cell cycle regulators in previous screens, spen (poc; Staehling-Hampton et al. 1999; Lane et al. 2000) and net (I. Harriharan, personal communication), both complemented 2.1 alleles. Taken together these data suggest that lgl corresponds to 2.1. To confirm that a lesion in lgl suppresses the DmcycE^JP phenotype, a null allele of lgl (lgl⁴) was tested for suppression of DmcycE^JP. However, lgl⁴ did not suppress the S-phase defect or the rough eye phenotype of DmcycE^JP to the same extent as 2.1 alleles did (Figure 2; and data not shown). However, halving the dosage of 2.1 alleles resulted in a greater increase in Cyclin E protein levels in DmcycE^JP eye discs than halving the dosage of lgl⁴/+ (Figure 3). It is possible that additional mutations in the lgl⁴ background may account for its poorer ability to dominantly suppress DmcycE^JP compared with 2.1 alleles. Consistent with lgl mutations being responsible for the suppression of DmcycE^JP, lgl-2.1 and other lgl mutant clones in the eye imaginal disc showed ectopic expression of Cyclin E, which could be suppressed by expression of lgl using a UAS-lgl transgene (N. Amin, A. Brumby, J. Secombe and H. Richardson, unpublished results).

Figure 2.—

Scanning-electron micrographs of adult eyes and BrdU labeling of eye imaginal discs from lgl, scrib, or dlg heterozygotes in a DmcycE^JP background. Genotypes are as indicated: DmcycE^JP; 2.1-23S9/+, DmcycE^JP; 2.1-27S3/+, DmcycE^JP; 2.1-E2S31/+, DmcycE^JP; 2.1-E6S2/+, DmcycE^JP; DmcycE^JP; scrib-63S15/+; DmcycE^JP; scrib¹/+; and dlg⁶/+; DmcycE^JP.

Figure 3.—

Cyclin E protein levels in eye imaginal discs from third instar larvae. Genotypes are as indicated: wild type (WT); DmcycE^JP; 2.1-23S9/+, DmcycE^JP; 2.1-E6S2/+, DmcycE^JP; lgl⁴/+, DmcycE^JP; and DmcycE^JP; scrib¹/+.

63S15 (scribble):

63S15 was localized to 97B–97D2 (Table 4), and consistent with a suppressor mapping in this region, Df(3R)T1^P, which failed to complement 63S15, was identified as a suppressor of DmcycE^JP in the screen of third chromosome deficiencies (Table 1). Cytological analysis of 63S15 showed that there was a lesion in the 97D region involving a translocation to the second chromosome (data not shown). By crosses to P alleles in the region, 63S15 was found to be allelic to l(3)j7b3, which is located in the first intron of a gene now known as scribble (Bilder and Perrimon 2000).

Scribble is a four-PDZ95-Dlg-ZO1 and multi-leucine-rich repeat containing protein localized to septate junctions and required for apical-basal polarity (Bilder and Perrimon 2000; Humbert et al. 2003). When homozygous, 63S15, like scribble null alleles, arrest as giant overgrown larvae due to amorphous overgrowth of imaginal discs and brain lobes, which is characteristic of neoplastic tumor suppressor mutants (Gateff et al. 1996; de Lorenzo et al. 1999; Bilder 2001). To confirm that lesions in scribble suppress DmcycE^JP, the l(3)j7b3 allele and stronger EMS alleles of scrib, scrib¹ and scrib² (Bilder and Perrimon 2000), were crossed into a DmcycE^JP background. The weak P allele, l(3)j7b3, did not suppress DmcycE^JP, although mild suppression was observed with scrib¹ and scrib² alleles, but not as well as with 63S15 (Figure 2 and data not shown). This suggests that 63S15 may be a stronger scribble allele than scrib¹ or scrib². In confirmation that scribble alleles suppress the DmcycE^JP phenotype, halving the dosage of scribble in DmcycE^JP eye discs leads to higher levels of Cyclin E (Figure 3 and data not shown) and scrib¹ and scrib² eye imaginal disc clones show ectopic expression of Cyclin E (Brumby and Richardson 2003).

lgl and scribble are neoplastic tumor suppressor genes that together with discs-large (dlg) act in the same pathway to regulate apical-basal cell polarity (Bilder et al. 2000; Humbert et al. 2003). Because of this function, we have termed these proteins cytoarchitectural tumor suppressors to highlight their role in cell structure. Consistent with this pathway being important in regulation of G1- to S-phase progression, a deficiency removing dlg, Df(1)v^N48, as well as a specific dlg allele (dlg⁶) showed suppression of DmcycE^JP (Tables 1 and 2; Figure 2). Scribble, Dlg, and Lgl have been recently shown to act antagonistically to the Crumbs cell polarity complex (Bilder et al. 2003; Tanentzapf and Tepass 2003), and consistent with this, a deficiency removing crumbs and a crumbs allele (crb²) dominantly enhanced DmcycE^JP (Table 1).

Scribble-interacting genes:

To determine whether a common pathway is involved in the mechanism by which the DmcycE^JP suppressors lead to deregulation of cell proliferation, we analyzed weak scribble mutant combinations for a dominant genetic interaction with other genes identified in the DmcycE genetic screen (Table 6). The trans-heterozygous combination of scrib⁵/scrib^l(3)jB709 or scrib¹/scrib⁵ results in adults with eye, bristle, and thorax-closure defects (not shown). Reducing the dose of the lgl (27S3, E2S31, and lgl⁴) showed a strong genetic interaction with the weak scrib allele phenotype, resulting in no scrib mutant progeny heterozygous for lgl. This is consistent with the previous observations that scribble mutations exhibit strong genetic interactions with lgl and dlg in the embryo (Bilder et al. 2000). Strikingly, halving the dosage of several other suppressor genes identified in the screen also resulted in very low numbers of scrib mutant progeny, most notably with dEB1 (2.5), phyl (2.2), the αPS3 integrin gene scab (2.11), the Brahma complex gene moira, and to a lesser extent brahma, as well as the unidentified 2.3, 2.4, and 2.9 genes. The mechanism of these interactions requires further analysis and relies on identifying the 2.3, 2.4, and 2.9 genes.

TABLE 6.

Genetic interactions withscribble—effect of halving the dosage of otherS(DmcycE^JP) genes on the viability of hypomorphicscribble allele combinations

		Total no. of progeny
Mutant allele		x−/+ or CyO/+; scrib1or5/TM6B	x−/+; scrib1/scrib5	% of expected no. of x−/+ scrib1/scrib5 progenya
2.1 (lgl)	27S3	304	0	0
E2S31	145	0	0
lgl4	248	0	0
2.2 (phyl)	28S2	231	0	0
2.3	65S12	227	0	0
36Fd	343	1	2
2.4	57S6	235	0	0
59S3	210	8	25
2.5 (dEB1)	42S11	147	3	13
2.9	E1S4	256	0	0
25S11	242	0	0
2.11 (scab)	E(sev-cycE)e93	86	1	8
Total no. of progeny
Mutant allele		x− scribJ7b3or5 /TM6B	x− scribJ7b3/scrib5	% of expected no. of x−/+ scrib1/scrib5 progenya
brm	2	401	47	24
mor	1	323	13	8
3.1	19S5	276	91	66
	24SX	279	58	42

^aThe expected number of scrib¹/scrib⁵ progeny = total TM6B progeny/4 × 0.61 was derived as follows: The control cross of scrib¹/TM6B × scrib⁵/TM6B gave 127 scrib¹/scrib⁵ out of 627 total progeny, i.e., 61% of the expected number. To control for this, the percentage of the expected number of x⁻/+ scrib¹/scrib⁵ progeny has been adjusted by this factor. The weaker scrib allele combination, −/+ scrib^J7b3/scrib⁵, gave expected Mendelian numbers of progeny. The expected number of scrib^J7b3/scrib⁵ progeny = total number of TM6B progeny/2.

The unidentified suppressors:

The map positions for suppressor groups 2.3, 2.4, 2.7, 2.8, 2.9, 2.10, 2.13, 3.1, 3.2, and 3.4 (1S2) and the single alleles 1S3, 13S1, 59S9, 59S18, and 68S10 were defined by genetic and deficiency mapping (Table 5). For the third chromosome suppressors, 3.1, 3.2, and 3.4 (1S2), the location of a suppressor within the defined region could be confirmed since the corresponding deficiencies dominantly suppressed DmcycE^JP (Table 1; and data not shown). However, for 13S1, 59S9, 59S18, and 68S10, the deficiencies that failed to complement these suppressors did not suppress DmcycE^JP (Table 1; and data not shown). For most of the unidentified suppressors complementation tests of all likely mutations and P alleles within the respective regions and Southern analysis of candidates have so far failed to identify the affected gene (Table 7); therefore, these suppressor mutations affect novel genes, which will require further analysis to identify. The exception is 2.3, where there are two candidates (Table 7 and see below). Potential candidates, with links to identified DmcycE^JP suppressors and thereby G1-S regulation, were found for many of the unidentified suppressors (see Table 7). Some of these candidates have been tested by complementation tests or Southern analysis and have been ruled out as being affected by the suppressor mutation (Table 7). Details on mapping and potential candidates for 2.3, 3.1, 1S3, and 59S9 are described below. For the details on other unidentified suppressors, see Tables 5 and 7.

TABLE 7.

Candidates for the unidentified suppressors

Suppressor group	Candidate gene	Gene function	Possible links to cell cycle regulation
2.3	l(2)36Fd?	Unknown	Unknown
hamlet?	Transcription factor in Dendrite morphogenesis (Mooreet al. 2002)	May be involved in regulating cytoskeletal, cell adhesion or cytoarchitectural tumor suppressor genes
2.4	Act42A	Actin 42A	Brahma complex component (Papoulas et al. 1998)
CG12792	WD40 domain	May be involved in proteolysis as is Cdc4/Ago (Moberg et al. 2001)
CG10412	Dbl-related (RhoGEF)	Activator of Rho family proteins and may regulate Rac (Blanchard 2000)
2.7	CG2727 (Emp)a	CD36-like	CD36-like proteins encode cell surface signaling
CG2736a	CD36-like	proteins that may have a role in adhesion and
CG3829a	CD36-like	signaling pathways regulating cell proliferation
(Greenwalt et al. 1992)
CG3770	Claudin-like	A tight junction protein involved in cell-cell adhesion and may have a role in inhibiting cell proliferation (Michl et al. 2003; Tepass et al. 2001)
2.8	wee1b	Cdc2 inhibitor (Campbell et al. 1995)	In combination with Cyclin A can drive entry into S phases (Dong et al. 1997; Thomas et al. 1997)
neuroligin	Cell adhesion protein	The mammalian homolog binds to Dlg4 (Bolliger et al. 2001)
2.9	Jitterbug (Filamin)a	Actin-binding protein	Possibly acts to regulate Rho family members (Sokol and Cooley 2003; Stossel et al. 2001)
moa	Cell adhesion	Possible role in Integrin signalling (Prout et al. 1997; Walsh and Brown 1998)
2.10	p120-catenina	Adherens junction component	Binds to E-cadherin and regulates Rho in mammalian cells (Aznar and Lacal 2001; Blanchard 2000; Jaffe and Hall 2002)
Gprk-1	G-protein-coupled receptor protein kinase with a RGS domain	Negative regulator of heterotrimeric G proteins, responsible for the rapid turnoff of G-protein-coupled receptor signaling pathways (De Vries and Gist Farquhar 1999)
2.13	CG2727 (Emp)	CD36-like	(See 2.7)
CG2736	CD36-like
CG3829	CD36-like
CG3770	Claudin-like
3.1	CG6445a	Cadherin-like	Possible role at adherens junctions
CG3885	Sec3-like exocyst component	Involved in docking at the plasma membrane, which is a function that Lgl has also been implicated in Lehman et al. (1999) and Musch et al. (2002)
3.2	CG6190	Ubiquitin ligase - HECT domain protein	A HECT domain ubiquitin ligase gene related to hyd, a tumor suppressor (de Lorenzo et al. 1999; Gateff et al. 1996)
3.4	CG5263 (smg)	Translational repressor	Role in neural cells (Clark et al. 2002); possible role in G1-S regulation, given the identification of RpS6 in the screen
1S3	APC1b	Adenomatous polyposis coli tumor suppressor	Possible role at adherens junction and may function with EB1 (Ahmed et al. 1998; Lu et al. 2001)
pins (rapsinoid)	Asymmetric division of neuroblasts	Directly interacts with Dlg (Bellaiche et al. 2001; Parmentier et al. 2000)
13S1	CG5555	Ring finger domain	Homologous to a protein shown to interact with the BRCA1 tumor suppressor protein in mammals (Sharan and Bradley 1998)
59S9	Spinophilin (neurabin)	Actin-binding scaffold protein	In mammalian cells is involved in binding to and upregulating Rac and p70-S6K activity (Buchsbaum et al. 2003)
59S18	draper	Protein with multiple extracellular EGF repeats, similar to laminin γ3 and Notch	Predicted to be involved in cell adhesion and signaling and involved in differentiation of neural cells (Egger et al. 2002)
68S10	canoeb	Component of the Adherens junction	Acts antagonistically to the Ras signaling pathway (Matsuo et al. 1997)
	CG12591	Ig C2-domain—cell adhesion	Possible role at adherens junction or in signaling (Tepass et al. 2001)

?, failed to complement 2.3 alleles, but testing did not confirm whether the mutations dominantly suppress DmcycE^JP.

^aTested by Southern analysis and no obvious disruptions were observed.

^bMutants were tested and shown to complement the DmcycE^JP suppressor.

2.3. (59S16, 65S12) location (36F7–37B8):

While 2.3^59S16 carries a deletion removing at least six complementation groups within the 36F7–37B8 region, including l(2)36Fd and l(2)37Ac, 2.3^65S12 was found to be lethal over the unidentified lethal gene l(2)36Fd, but gave ∼5% escapers over l(2)37Ac. 2.3^65S12 is therefore likely to be a smaller lesion affecting both of these uncharacterized genes. A recently characterized gene in the 36F region, hamlet, which is a transcription factor involved in dendrite morphogenesis (Moore et al. 2002), was also tested for allelism with 2.3 and failed to complement 2.3^59S16 and 2.3^65S12 but not l(2)36Fd. Further analysis is required to determine whether hamlet or l(2)36Fd corresponds to the 2.3 suppressor.

3.1. (19S5, 24SX, 58S5, 62S2) location [73D–74F (74B1–74C1)]:

Consistent with the map position defined by deficiency mapping, chromosome cytology revealed that 58S5 contained a deletion in the 74A–F region, and it failed to complement several lethal alleles in the region. The cadherin-like gene, CG6445 (Cad74A), was considered a candidate, since the cadherin-like protein, Fat, is a tumor suppressor in Drosophila (Gateff et al. 1996; de Lorenzo et al. 1999). Southern analysis failed to reveal any alterations in this gene in 3.1 alleles (data not shown). The method of male recombination (Svoboda et al. 1995) was then used to further define the map position of the 3.1 alleles, 19S5 and 24S10 relative to several P alleles, revealing that the lethal associated with 3.1 mapped to the right of blot (74B1–2) and to the left of l(3)S070006 (allelic to l(3)L6750 = frc at 74B4), l(3)00073 (74C1–2), and EIP74EF (74D2–5). Taken together these data suggest that 3.1 maps between 74B1 and 74B4. A candidate gene within this region, CG3885, encodes a Sec3-like protein, a component of the exocyst complex involved in docking at the plasma membrane, which is a function that Lgl has also been implicated in (Lehman et al. 1999; Musch et al. 2002).

1S3 location [98A–100B (98A5–98E3)]:

Chromosome cytology showed that 1S3 contained a translocation breakpoint at 98C (data not shown). Since there is a hole in the deficiency collection between 98A5 and 98E3, it is likely that 1S3 maps within this region. A candidate in the 98A5–98E3 region was APC1 (encoding the Adenomatous polyposis tumor suppressor; Ahmed et al. 1998); however, mutations in APC1 (APC^Q8 and APC^X1) complemented 1S3. Another candidate is raps (pins), which encodes a protein involved in asymmetric division of neuroblasts and directly interacts with Dlg (Parmentier et al. 2000; Bellaiche et al. 2001). Further analysis is required to test whether raps mutations are allelic to 1S3.

59S9 location (62D2–62F5):

Consistent with this location for 59S9, cytological analysis revealed a breakpoint at 62B. A possible candidate in this region is spinophilin (neurabin), encoding an actin-binding scaffold protein, which in mammalian cells is involved in binding to and upregulating Rac and p70-S6K activity (Buchsbaum et al. 2003). Since another gene involved in Rac activation, trio, was identified as a suppressor of DmcycE^JP it is possible that spinophilin is also a suppressor. Furthermore, Drosophila mutations in spinophilin are semilethal (Keegan et al. 2001), as is 59S9.

Further analysis is needed to investigate whether the potential candidates for these suppressors, listed above and in Table 7, are disrupted by the suppressor mutations and for the identification of the suppressors.

DISCUSSION

In this study, we have identified genetic interactors of cyclin E by screening deficiencies, by testing candidate genes, and through EMS and X-ray mutagenesis screens. This work has led to the identification of many genes that when mutated have the ability to dominantly modify the DmcycE^JP adult rough eye phenotype and S-phase defect in third instar larval eye imaginal discs. In addition to genes already known to be regulators of Drosophila cyclin E or G1-S progression, such as E2F1; retinoblastoma (Rbf); ago (cdc4) encoding a protein involved in Cyclin E degradation (Moberg et al. 2001); the EGF receptor pathway genes Egfr and Ras85D, which act to promote Cyclin E protein accumulation (Prober and Edgar 2000; Brumby and Richardson 2003); and Hh signaling pathway genes, which act to promote cyclin E transcription (Duman-Scheel et al. 2002); this screen led to the identification of many novel cyclin E interactors. This study has mainly concentrated on the suppressors of DmcycE^JP, although from the deficiency screen and specifically testing candidates, we identified axin (an inhibitor of Wg signaling), rho1, and crumbs as enhancers of DmcycE^JP, which therefore may act as novel positive regulators of G1-S progression. The suppressors of DmcycE^JP identified include the following classes: (1) chromatin remodeling genes brm, mor, Trl, or the transcription factor Zn72D; (2) signaling pathway genes phyl, sina, trio, Abl, RpS6, wg and Wg pathway effectors dsh and arm; (3) genes encoding cytoskeletal proteins dEB1 (encoding a microtubule-binding protein) and expanded (encoding a FERM domain cytoskeletal protein and hyperplastic tumor suppressor); (4) genes encoding cell adhesion proteins scab (encoding an α-integrin), cadN (N-Cadherin), shg (E-Cadherin), and fat (encoding an atypical-cadherin and hyperplastic tumor suppressor); and (5) cytoarchitectural tumor suppressor genes scribble, lgl, and dlg, required for apical-basal cell polarity and cell proliferation inhibition. While some of these genes (brm, mor, expanded, fat, scribble, and lgl) have been previously shown or implicated to play a role in negatively regulating G1-S (Gateff et al. 1996; de Lorenzo et al. 1999; Staehling-Hampton et al. 1999; Bilder et al. 2000), a potential role for Trl, Znf72D, phyl, sina, trio, Abl, RpS6, wg, dsh, arm, dEB1, scab, cadN, and shg in inhibiting G1-S progression in Drosophila is novel. Further studies are required to determine whether Abl, RpS6, wg, dsh, arm, and shg do indeed suppress DmcycE^JP by acting at the S-phase level and to understand the mechanism by which these genes act in G1-S regulation. The identification of novel classes of presumptive negative regulators of cyclin E or G1-S progression highlights the power of Drosophila whole-animal genetics as a tool for revealing new cell proliferation pathways.

It is unclear at present how many of the DmcycE^JP modifiers identified in our screen bear upon the role of Cyclin E in DNA replication or centrosome duplication (see Introduction). Brahma and Moira are likely to be downstream targets of Cyclin E/cdk2 that may impact upon transcriptional regulation or DNA replication (Brumby et al. 2002), but whether other interactors act upstream or downstream of Cyclin E remains to be determined. The only cyclin E interactor we identified that has been shown to be associated with the centrosome is EB1 (Rehberg and Graf 2002); however, whether this reflects upon the role for Cyclin E in centrosome duplication in Drosophila is unclear. A recent study has shown that the Drosophila SkpA, a component of SCF ubiquitin ligases, regulates centrosome duplication independently of Cyclin E accumulation (Murphy 2003).

Similar genetic screens carried out using phenotypes generated by overexpression of cyclin E (Lane et al. 2000) or the G1-S regulators E2F1/Dp (Staehling-Hampton et al. 1999), Rbf (Duman-Scheel et al. 2002), and human p21 (Cdk2 inhibitor; I. Hariharan, personal communication) have revealed a more restricted set of interacting genes than that obtained in our cyclin E hypomorphic allele genetic screen. The GMR-E2F1/Dp screen (Staehling-Hampton et al. 1999) revealed alleles of the chromatin remodeling genes brm, mor, and osa and of the transcription factor pointed, an effector of the Egfr-Ras signaling pathway, as enhancers. This is consistent with our identification of brm and mor as suppressors of the hypomorphic cyclin E phenotype in our mutagenesis screen. In addition, we tested alleles of osa and showed that they suppressed the hypomorphic cyclin E phenotype (Brumby et al. 2002). The sevenless-cyclin E screen (Lane et al. 2000) revealed alleles in identified cell cycle genes cdk2 (as a suppressor), dacapo (as an enhancer), and E2F1 (a suspected gain-of-function allele as an enhancer) and identified as an enhancer the novel gene spen (poc), also identified in the GMR-E2F1/Dp screen (Staehling-Hampton et al. 1999). Spen (Poc) is a RNP-type RNA-binding protein that has recently been shown to be required for Wg signaling in imaginal discs (Lin et al. 2003). We have not identified spen (poc) as a suppressor in our genetic screen, but alleles of spen (poc) were tested and shown to suppress DmcycE^JP (Table 2), consistent with the Wg signaling pathway acting to negatively regulate G1-S progression in the eye disc. As detailed above, we have shown that one of the single alleles identified as an enhancer in the sevenless-cyclin E screen is allelic to our DmcycE^JP suppressor 2.11, which we have identified as scab. In the GMR-Rbf screen, alleles of patched, encoding an inhibitor of Hedgehog (Hh) signaling, were identified as dominant suppressors (Duman-Scheel et al. 2002). Although our mutagenesis screen did not reveal alleles of patched, patched alleles strongly suppressed DmcycE^JP (Table 2), consistent with the notion that Hh signaling leads to increased transcription of cyclin E (Duman-Scheel et al. 2002). The greater number of interactors that we obtained in our screen may be due to the fact that our screen was of a cyclin E hypomorphic phenotype that affected cell proliferation in early eye development as well as the post-MF S phases and may therefore have been more sensitive to gene dosage than the overexpression screens. Furthermore, unlike the overexpression screens, the cyclin E hypomorphic screen is more likely to reveal genes that are upstream of cyclin E expression.

The DmcycE^JP suppressor genes we have identified from our mutagenic screen are mostly distinct from Drosophila tumor suppressors previously described (Torok et al. 1993; Gateff et al. 1996; de Lorenzo et al. 1999). Recently, clonal screens have revealed a novel pathway involved in inhibiting G1-S progression and cell death in the Drosophila eye (Hay and Guo 2003). This pathway includes lats (warts), salvador, and hippo, and although this pathway has been recently shown to regulate cyclin E at possibly both a transcriptional and protein stability level, we did not identify alleles of these genes in our genetic screen. Alleles of hippo, at least, have been shown to suppress the DmcycE^JP phenotype (Wu et al. 2003). The fact that we did not identify hippo in our mutagenesis screen may have been because the screen was not saturating. However, lats (warts) alleles did not show appreciable suppression of DmcycE^JP (Table 2); therefore it is possible that only certain mutations of this pathway are capable of dominant suppression.

Also pertinent to our study is the recent Drosophila protein interaction map determined by yeast two-hybrid analyses (Giot et al. 2003). None of our identified cyclin E genetic interactors were identical to the 15 interactors identified by the protein interaction study (Giot et al. 2003), but many proteins identified in our screen were not analyzed in their screen (e.g., Brahma, Moira, Scab, CadN, Dsh, Scribble, Crumbs, Expanded, and Abl). Most of the 15 yeast two-hybrid interactors with Cyclin E are uncharacterized, but of the characterized proteins, Combgap, a transcription factor, has been implicated in cell proliferation via its effect on Ci expression (Campbell and Tomlinson 2000). Of the other characterized interactors, Gliolectin is involved in cell adhesion in axon pathfinding (Sharrow and Tiemeyer 2001) and Traf2 is involved in Dorsal activation (Shen et al. 2001), but no cell proliferation role has been described for these proteins. Some of the Cyclin E yeast two-hybrid interacting genes map to regions where cyclin E genetic interactors have been mapped (not shown) and are candidates for future analysis.

Whether the genetic suppressors of cyclin E identified in our screen can all be connected in a common pathway or represent several converging pathways acting upon G1-S progression in the eye imaginal disc remains to be determined. As a first step to explore this we examined interactions between a weak scrib mutant and S(DmcycE^JP) alleles, which revealed genetic interactions with lgl, phyl, dEB1 scab, mor, the unidentified suppressors 2.3, 2.4, and 2.9, and to a lesser extent brm. This analysis provides a connection between chromatin remodeling, signaling, cytoskeletal, cell-cell adhesion, and cytoarchitectural suppressor genes. How exactly these pathways may be connected and whether other genes identified in the DmcycE^JP screen are also functionally connected now warrant further investigation.

Interestingly, many of the genes identified in the screen have roles in cell polarity; for example, scrib, dlg, lgl, and crumbs are involved in apical-basal cell polarity, while dlg, fat, expanded, and the Wg pathway, via Rho and Jnk, have roles in planar polarity (Blaumueller and Mlodzik 2000; Bellaiche et al. 2001; Yang et al. 2002; Eaton 2003; Fanto et al. 2003). Moreover, E-cadherin (shg) and β-catenin (arm) function at the adherens junction, which is important in both apical-basal cell polarity and cell-cell adhesion (Tepass et al. 2001). Whether other cell polarity genes, such as bazooka, par3, apkc, patj, and stardust (Humbert et al. 2003), are also DmcycE^JP modifiers and the molecular mechanism by which this occurs require further analysis. Pertinent to this, a recent study has shown that apkc clones have reduced cell division and that apkc mutants can suppress the overgrowth of lgl mutants, suggesting that upregulation of apkc contributes to the overgrowth phenotype of lgl, and perhaps also scrib and dlg, mutants (Rolls et al. 2003).

How are junctional components connected to signaling pathways or to the cell cycle machinery? In mammalian cells, the Frizzled receptors, Fz1, Fz2, Fz4, and Fz7, have been shown to bind to mammalian Dlg1 (Hering and Sheng 2002), which may therefore provide a connection between apical-basal and the Frizzled-Rho-Jnk planar polarity pathway (Adler and Lee 2001), as well as to the canonical Wg-Arm (β-catenin) pathway to effect S-phase entry (Figure 4). Furthermore, mammalian scrib genetically and physically interacts with the planar polarity gene, vang (strabismus) (Kallay et al. 2003; Montcouquiol et al. 2003; Murdoch et al. 2003). Mammalian Vang is a potential tumor suppressor that can act to regulate the Wg-Arm pathway (Katoh 2002). If Vang acts similarly in Drosophila, it would provide another connection between planar polarity, apical-basal polarity, and Wg signaling pathways. Connections between polarity proteins and the Egfr signaling pathway have also been observed in Caenorhabditis elegans and mammalian cells (Simske et al. 1996; Huang et al. 2003). Consistent with this, antagonistic interactions between E-cadherin in adherens junction function and the Egfr signaling pathway have been observed in the Drosophila nervous system (Dumstrei et al. 2002), and if this also occurs in the eye imaginal disc then decreasing E-cadherin levels would be expected to cause an increase in Egfr-Ras signaling that would lead to increased Cyclin E protein (Brumby and Richardson 2003). Furthermore, there is evidence that the FERM domain protein Expanded, which functions together with another FERM domain protein, Merlin, a homolog of the NF2 tumor suppressor, modulates the Dpp signaling pathway (McCartney et al. 2000) and in mammalian cells NF2 can inhibit Ras signaling (Lim et al. 2003). There is also a precedent for a connection between Integrin signaling and cell polarity pathways, since the transmembrane Laminin receptor Dystroglycan has been shown to have a role in epithelial cell apical-basal polarity (Deng et al. 2003). This now raises the question of whether Scab (αPS3-Integrin) plays a role in apical-basal cell polarity. In mammalian cells, integrins act via focal adhesion kinase (Fak) to activate Rho-family GTPases (Schoenwaelder and Burridge 1999) and recently it has been reported that integrins are important for the localization of aPKC (Datta et al. 2003). In Drosophila, a role for scab in modulation of Dpp signaling has been described in wing vein formation (Araujo et al. 2003), suggesting a mechanism by which scab may also affect cell proliferation. Furthermore, there is a connection between the Trio-Rac-Abl pathway and polarity, since Trio interacts with the Lar receptor-like tyrosine phosphatase, which has recently been shown to have a role in epithelial planar polarity (Frydman and Spradling 2001).

Figure 4.—

Possible pathways connecting cyclin E-interacting genes. Interactors identified in our cyclin E screen are shaded. Direct protein interactions between cyclin E interactors or other relevant proteins are indicated by the double-headed arrows. Arrows indicate positive interactions while barred lines indicate negative interactions. Not shown are interactions between Scab and the Dpp pathway, between E-cadherin and the Egfr pathway, between Fat and Atrophin (a nuclear corepressor), and between Expanded/Merlin and the Dpp and Egfr pathways. *, genes that genetically interact with scribble. See the text for details.

The Drosophila microtubule-binding protein dEB1 has also been implicated in playing a role in adherens junction function and cell polarity by RNA ablation studies (Lu et al. 2001; Rogers et al. 2002). Interestingly, the recently published study on Drosophila protein interactions using yeast two-hybrid analysis has revealed that dEB1 binds to the Sina homolog CG13030, providing a connection to the Sina-Phyl pathway (Giot et al. 2003). Sina and the Sina homolog also bind to Rasputin (Rin), a homolog of the RasGAP-binding protein G3BP, which has a role in planar polarity via effects on the Rho signaling pathway (Pazman et al. 2000). Thus the Sina-Phyl complex may act via Rasputin to negatively regulate Ras and Rho signaling and thereby G1-S progression (Figure 4). The protein interaction study (Giot et al. 2003) has also revealed that RpS6, identified as a suppressor in our screen, binds to the planar polarity protein Vang/Strabismus, which was not tested in our screen. Interestingly in mammalian cells, Cdc42, a Rho-family GTPase component of the apical Par6 complex, functions via p70-S6 kinase to upregulate cyclin E transcription (Chou et al. 2003) and disruption of RpS6 in mice results in a specific block in cyclin E expression (Volarevic et al. 2000). The yeast two-hybrid analysis study (Giot et al. 2003) also revealed protein interactions between Zn72D and Actin 5C, a component of the Brahma complex (Papoulas et al. 1998), between the Brahma-associated protein Bap60 and the apical zone polarity protein aPKC (Humbert et al. 2003), and between Dlg or Lgl and zinc finger transcription factors. There are precedents for functional interactions between cell polarity proteins and nuclear corepressors, for example, between Drosophila Fat (atypical cadherin involved in planar polarity) and Atrophin (Fanto et al. 2003), suggesting that yeast two-hybrid interactions between the Brahma complex or the zinc finger transcription factors and cell polarity proteins may be functionally relevant, although further investigation is required. Although there may be many pathways that connect the Cyclin E interactors identified in this screen to G1-S progression, the examples above suggest ways in which cell polarity proteins may link to signaling pathways or directly to chromatin remodeling, corepressors, or transcription factors to regulate cyclin E or the transcription of other G1- to S-phase genes (Figure 4).

In summary, the identification in our cyclin E screen of genes that were not necessarily predicted to play roles in G1-S progression highlights the importance of using whole-animal genetics to investigate G1-S regulation. The identified cyclin E genetic suppressors are conserved in mammals and given their demonstrated or presumptive roles as inhibitors of G1-S progression in Drosophila are candidates for tumor suppressors in mammalian cancers.