Negative-feedback regulation of proneural proteins controls the timing of neural precursor division

Proneural proteins of basic helix-loop-helix (bHLH) transcription factors are master controllers that initiate and execute neurogenic programs. Proneural proteins are deployed in different lineages to generate a vast array of neuronal types. Thus, precise controls of proneural protein activities,both positive and negative, are crucial in building the nervous system (for reviews, see Hassan and Bellen,2000; Bertrand et al.,2002; Kiefer et al.,2005).

Achaete (Ac) and Scute (Sc) are the first identified members of the bHLH proneural protein family, and are required for formation of Drosophila external sensory (ES) organs(Villares and Cabrera, 1987). Expression of Ac and Sc in ectodermal cells endows these cells with the ability to develop into sensory organ precursors (SOPs), which then undergo asymmetric cell divisions to generate distinct daughter cells that constitute an ES organ. In mutants that remove both ac and sc loci, no ES organ is formed owing to the failure of SOP generation. As master regulators, misexpression of either Ac or Sc in ectodermal cells, in which they auto- and crossregulate each other, induces SOP formation, leading to the generation of ectopic ES organs (Campuzano et al., 1985; Romani et al.,1989; Rodriguez et al.,1990; Cubas et al.,1991; Skeath and Carroll,1991; Usui and Kimura,1992).

The Ac and Sc proneural proteins induce a developmental program for ES organs by activating transcription of an array of target genes(Reeves and Posakony, 2005). One of the target genes is phyl. phyl is necessary for SOP specification (Pi et al.,2001), and its expression in SOPs is activated through E-box motifs, binding sites for bHLH proneural proteins(Pi et al., 2004). Phyl functions as the substrate adaptor of an E3 ubiquitin ligase complex that includes the RING-finger protein Seven-in-absentia (Sina)(Li et al., 2002; Cooper et al., 2007). Binding of Phyl to Tramtrack (Ttk), a transcriptional repressor that inhibits neuronal potential, results in degradation of Ttk in a sina-dependent manner(Li et al., 1997; Tang et al., 1997; Pi et al., 2001; Badenhorst et al., 2002; Cooper et al., 2007). Thus,one mechanism underlying SOP specification is through the Ac- and Sc-activated Phyl/Sina degradation machinery that relieves the Ttk transcriptional repression.

Following specification, SOPs divide asymmetrically to generate pIIa and pIIb, which divide again and eventually give rise to distinct daughter cells of an ES organ (Hartenstein and Posakony,1989; Gho et al.,1999; Reddy and Rodrigues,1999; Fichelson and Gho,2003). SOPs are specified during the G2 phase(Usui and Kimura, 1992; Kimura et al., 1997) and the timings of subsequent G2-M transition play crucial roles in SOP and daughter cell fate specification. When SOP division is delayed by downregulating the activity of the cyclin-dependent kinase Cdc2, undivided SOPs adopt the pIIb fate, forming an abnormal ES organ(Fichelson and Gho, 2004). However, when SOPs are forced to enter mitosis by misexpression of String(Stg), the Drosophila homolog of the mitotic inducer Cdc25, SOP specification is defective, leading to loss of ES organs(Kimura et al., 1997; Negre et al., 2003). Thus, the regulatory circuitry of SOP division must incorporate a crosstalk mechanism with SOP specification to ensure that SOPs divide at the correct time.

Although much is known about the molecular mechanisms in SOP specification(Pi and Chien, 2007) and the asymmetric division of SOP (Yu et al.,2006), little is known about the process in between them: how the timing of G2-M transition is regulated by prerequisite specification in the G2 phase. It has been long speculated that proneural proteins play a role in this process. In wing imaginal disks, where misexpression of sc in the proliferating cells is sufficient to reduce stg expression and arrest cells in the G2 phase, loss of ac and sc functions re-activates stg expression in the dorsal and ventral subdomains of the zone of non-proliferating cells(Johnston and Edgar, 1998). Sc expression in SOPs is extinguished just before entry into mitosis(Romani et al., 1989; Skeath and Carroll, 1991; Cubas et al., 1991). Thus,although proneural protein levels peak in the G2 phase during SOP specification, they are depleted prior to the G2-M transition. The molecular mechanism underlying the proneural protein downregulation and the functional significance of such downregulation to the timing of G2-M transition remain elusive.

We report here that in phyl mutants, division of SOPs is blocked or severely delayed at the G2-M transition, owing to lack of stgexpression. Concomitantly, Ac and Sc proteins accumulate in phylmutant SOPs. Although downregulation in SOPs before mitotic entry still occurs on constitutively expressed proneural proteins, it is blocked in proteasome mutants, suggesting that this downregulation process is mediated through proteasomal degradation. Phyl directly associates with proneural proteins in Drosophila cells, and functions as an adaptor between proneural protein and Sina, as shown by the yeast bridge assay. Finally, reduction in the ac and sc gene doses suppresses the SOP division defect in phyl mutants. Taken together, we propose that proneural proteins initiate their own degradation through ubiquitin E3 ligase complex Phyl/Sina to determine the timing of G2-M transition.

The following fly lines were used: w¹¹¹² (wild-type), phyl², phyl⁴(Dickson et al., 1995), UAS-myc-phyl (Pi et al.,2001), UAS-flag-phyl (this study), UAS-stg(Neufeld et al., 1998), UAS-myc-ac (this study), UAS-myc-sc (this study), UAS-DTS5 (Schweisguth,1999), Eq-Gal4 (Pi et al., 2001), ap-Gal4(Milan and Cohen, 1999), sc^10-1 (Villares and Cabrera, 1987), sc^M6(Gomez-Skarmeta et al., 1995), ac^cami (Marcellini et al., 2005), sina² and sina³ (Carthew and Rubin, 1990). phyl² or phyl-misexpression mitotic clones were generated in hs-FLP;FRT^G13 phyl²/FRT^G13 ubi-GPF or hs-FLP; act>Y⁺>GAL4 UAS-GFP/+; UAS-myc-phyl/+ pupae,respectively. sina³ mutant clones were generated in hs-FLP; FRT^2A sina³/FRT^2A ubi-GFP. For the stg rescue experiment, phyl²/phyl⁴;hs-Gal4 UAS-stg pupae of 16-18 hours APF were given the following heat shock treatment: 37°C for 30 minutes, 25°C for 60 minutes and 37°C for 30 minutes. Pupae were then placed at 25°C for another 2 or 4 hours before dissection.

For antibody staining of pupal thoraces, pupae were dissected in 1×PBS and fixed in fresh 4% paraformadehyde. After washing with 1×PBT (Triton X-100, 0.1%), thoraces were incubated with the following primary antibodies: anti-Sens (1:1000) (Nolo et al., 2000), anti-PH3 (1:100) (Upstate), anti-CycB (1:75) (Santa Cruitz), anti-Ac (1:10) (Hybridoma Bank), anti-Sc (1:50)(Skeath and Carroll, 1991),anti-Da (1:100) (Cronmiller and Cummings,1993), mouse anti-Myc (1: 500) (Santa Cruz) and anti-Hnt (1:25)(Hybridoma Bank).

The staged pupae were first dissected in 1×PBS, and then incubated in BrdU solution (0.1 mg/ml in Grace medium) for 40 minutes. After BrdU incorporation, thoraces were fixed in 4% paraformaldehyde for 45 minutes, and followed by HCl hydrolysis (3 N for regular treatment, 1.5 N for visualizing the GFP clone marker) for 15 minutes. Tissues were then stained with mouse anti-BrdU (1:20) (Becton Dickinson) overnight.

The same protocal described by Tautz and Pfeifle(Tautz and Pfeifle, 1989) was used for in situ hybridization. stg antisense RNA probe was generated using DIG RNA Labeling Kit (Sp6/T7) (Roche).

S2 cells (1×10⁷) were plated on a 10 cm dish and transfected with 5 μg of pUAST-myc-ac (or pUAS-myc-sc or pUAS-ha-da), 5 μg of pUAST-3flag-phyl or vector and 3μg pWAGAL4. The transfected cells were pretreated with proteasome inhibitors MG132 (50 mM) for 6 hours before collecting cells. Forty-eight hours after transfection, cells were washed twice with cold 1× TBS and then lysed in 500 μl mRIPA buffer [50 mM Tris-HCl (pH 7.8), 150 mM NaCl, 5 mM EDTA (pH 8.0), 0.5% Triton X-100, 0.5% NP-40 and complete protease inhibitor cocktail tablets (Roche)] on ice for 30 minutes. The lysate was centrifuged for 1 hour. The supernatant of the lysate was first pre-cleared with 40 μl of Protein-A/G-agarose beads (Calbiochem), followed by immunoprecipitation with 30 μl anti-FLAG M2 affinity gel (Sigma). After incubation for 2 hour at 4°C, the beads were collected by centrifugation and washed three times with 1 ml of mRIPA buffer.

GST-fusion proteins were expressed in BL21(DE3) cells (Novagen). After purification, 1 μg GST-fusion protein or 5 μg GST protein were incubated with S³⁵-labelled in vitro translated protein (TNT system by Promega) in HEMNK buffer containing 1.25 mg/ml BSA. After a 1 hour incubation at 4°C, GST-beads were washed three times with HEMEK buffer and samples were analyzed by SDS-PAGE.

For bridge assay, sina and phyl were cloned into the pBridge vector (Clonetech) to allow expression of GBD-Sina and Phyl proteins,respectively, in yeast. ac was cloned into pGADT7 (Clonetech) to allow expression of GAD-Ac. Plasmids were co-transformed into yeast strain AH109. Interaction was scored by comparing the growth on SD-Trp/-Leu,SD-Trp/-Leu/-His/-Ade and SD-Trp/-Leu/-His/-Ade/-Met plates.

Senseless (Sens) is expressed in all SOPs and SOP daughter cells during ES organ development (Nolo et al.,2000). All microchaetal SOPs in the pupal thorax were specified at 14 hours after puparium formation (APF), as shown by the expression of Sens(Fig. 1A). SOPs divide once to generate two Sens-positive cells, the pIIa and pIIb, at 14-16 hours APF(Fig. 1B), indicating that SOP division is under strict temporal control. In the hypomorphic phyl²/phyl⁴ mutants, one-third of SOPs are specified and more than 50% of these SOPs remain as single Sens-positive cells (Pi et al., 2004),suggesting that SOP division control is defective in phyl mutants. To investigate the SOP division defect further, we analyzed mutant clones homozygous for the phyl² null allele. Although SOPs were largely unspecified (45/66, 69%) in phyl² clones(Fig. 1C), the few SOPs that were specified, as shown by Sens expression, still remained as single cells at 22-24 hours APF (Fig. 1C). At the same stage, four or five Sens-positive daughter cells had been generated in the SOP lineage in the neighboring wild-type tissues(Fig. 1C, GFP-positive area). At 24-28 hours APF, 53% (nine out of 15) of the SOPs in phyl² mutant clones (n=6) had divided into two cells (Fig. 1D), whereas the rest still remained undivided, suggesting that SOP cell cycle progression is delayed or absent in phyl² mutants. phyl is specifically expressed in proneural clusters and SOPs(Pi et al., 2001; Reeves and Posakony, 2005) and the clone size of phyl² was comparable with that of its twin spot (data not shown), indicating that the cell cycle defect in SOPs is not due to a general defect in the proliferation of thoracic tissue.

To examine whether phyl regulates a specific cell cycle phase in SOP, phase-specific markers were used to stain the pupal thorax. At 13-15 hours APF, the presence of phosphorylated histone H3 (PH3), a mitotic marker,could be clearly detected in dividing wild-type SOPs(Fig. 1E). However, PH3 signal was never detected in phyl²-mutant SOPs at 14-16 hours APF(Fig. 1F) and most of the mutant SOPs (15/17) remained PH3-negative between 16 and 24 hours APF (data not shown). Cyclin B (CycB) accumulation starts in the S phase, and peaks at the late G2 phase (Baker and Yu,2001). Accumulation of CycB was detected in wild-type SOPs at 12-14 hours APF (Fig. 1G) and in phyl² mutant SOPs (17/17, n=7) during 16-24 hours APF (Fig. 1H, showing a representative staining at 16-18 hours APF), suggesting that these phyl² mutant SOPs remain in the G2 or S phase. To distinguish between these two possibilities, the BrdU-incorporation assay was performed. In wild-type pupal thorax at 12-14 hours APF, BrdU incorporation was never detected in single SOPs but could be detected in some daughter cells of divided SOPs (Fig. 1I). BrdU signal was not detected either in phyl²/phyl⁴hypomorphic mutant SOPs between 16 and 24 hours APF(Fig. 1J, showing a representative at 22-24 hours APF) or in phyl²-mutant SOPs between 18 and 24 hours APF (Fig. 1K, showing a representative at 20-22 hours APF), demonstrating that they are not arrested at S phase. These results together indicate that G2-M transition is blocked or delayed in phyl mutant SOPs.

Cdc25 induces mitosis by dephosphorylating the cyclin-dependent kinase Cdc2, leading to G2-M transition (Russell and Nurse, 1986; Edgar and O'Farrell, 1990; Sadhu et al.,1990). string (stg) encodes the Drosophila Cdc25 homolog and is expressed in premitotic cells during development (Lehman et al.,1999). To examine whether phyl promotes G2-M transition by controlling stg expression, we first analyzed the stgmRNA patterns in SOPs located at the anterior wing margin (AWM) because wing disks are more accessible to RNA in situ analyses than pupal thorax. Moreover,division of AWM SOPs was also delayed or absent in phyl mutants (data not shown). At 2-4 hours APF, reproducibly stg mRNA expression was detected in the AWM SOPs (Fig. 2A). stg expression in phyl²/phyl⁴ disks was almost abolished or significantly reduced (Fig. 2B). These results suggest that phyl is required for stg expression in SOPs, and that failure in G2-M transition of thoracic SOPs might be caused by the absence of stg expression. To test this, stg was misexpressed in phyl mutants to rescue the SOP division defect in the thorax. Although all wild-type SOPs had divided by 22 hours APF (Fig. 2D,showing representative staining pattern at 24-26 hours APF), only 15±4%and 24±8% of SOPs in phyl²/phyl⁴ mutants had divided at 22-24 and 24-26 hours APF, respectively (column 1 and 3 in Fig. 2C, and Fig. 2E). When stg was transiently induced in the phyl²/phyl⁴ mutant pupae by heat shock treatment at 18-20 hours APF (see Materials and methods),84±7% and 73±6% of SOPs had divided at 22-24 and 24-26 hours APF, respectively (column 2 and 4 in Fig. 2C, and Fig. 2F),demonstrating that cell division delay in phyl mutants is rescued by stg misexpression. Together, these data indicate that phylpromotes G2-M transition by positively regulating stg expression in SOPs.

In wing disks, proneural proteins Ac and Sc negatively regulate stg expression, leading to G2 arrest in the dorsal and ventral non-proliferating cells surrounding the wing margin(Johnston and Edgar, 1998). Gradually diminishing Ac and Sc protein levels in SOPs, therefore, could be a prerequisite for stg expression and the consequent G2-M transition. At 12-13 hours APF, Ac protein was enriched in newly specified SOPs that expressed low levels of Sens (arrowheads in Fig. 3A). Interestingly, we found that in some SOPs where Ac expression was diminishing, Sens expression reached higher levels, indicating that they are more mature SOPs (arrows). The Ac protein disappeared from the Sens-enriched late-stage SOPs at 13-14 hours APF (GFP-positive area in Fig. 3E). Similar timings of Sc expression in thoracic SOPs were also observed (Fig. 4A, and GFP-positive area in Fig. 4C). Thus, these data show that proneural protein levels are gradually diminishing during SOP specification.

Next, we sought to examine whether post-transcriptional mechanism is involved in the downregulation of proneural proteins by expressing the Myc-tagged Ac protein with a heterologous promoter. The Eq-Gal4driver induces a ubiquitous expression pattern in the prospective thoracic region of the wing discs (Pi et al.,2001). When the UAS-myc-ac transgene expression was activated by Eq-Gal4, ectopic SOPs were induced in the wing disks, as shown by Sens expression (arrowheads and arrows in Fig. 3B″). Although Myc-Ac was detected in newly specified SOPs (arrowheads in Fig. 3B) and in most non-SOP disk cells, significant downregulation of Myc-Ac protein was invariably observed in late-stage SOPs with stronger Sens expression (arrows in Fig. 3B). Downregulation of Myc-Sc protein in mature SOPs was also observed(Fig. 4B). Thus, the downregulation of ectopically expressed Ac and Sc proteins suggests the involvement of a post-transcriptional mechanism in late stages of SOPs prior to the cell division.

We then tested whether downregulation of proneural protein is a result of proteolysis by the 26S proteasome. To inactivate the proteasome, we used a dominant-negative, temperature-sensitive form of the proteasome β6 catalytic subunit, DTS5 (Schweisguth,1999). When DTS5 was misexpressed in the thoracic tissue by the apterous (ap)-Gal4(Milan and Cohen, 1999) at a restrictive temperature (29°C), endogenous Ac protein levels were highly elevated in SOPs (compare Fig. 3D with 3C), indicating that the proteasome degrades Ac in SOPs.

Thus, our results indicate that proneural proteins in SOPs are downregulated through post-transcriptional, proteasome-dependent mechanism before mitotic entry. As Phyl is a component of ubiquitin E3 ligase complex and is required for G2-M transition, it raises the possibility that phyl downregulates proneural protein levels before SOP division. Indeed, in phyl² mutant clones at 13-14 hours APF, Ac protein was still maintained at high levels(Fig. 3E,3E′), and lower levels of Sens were detected in these mutant SOPs when compared with that in wild-type SOPs (Fig. 3E″). Ac protein levels were also maintained in phyl²/phyl⁴ pupal thorax (data not shown). Upregulation of Sc protein levels in SOPs was also observed in phyl² mutant clones(Fig. 4C). Collectively, these results indicate that phyl is required for the downregulation of proneural proteins in SOPs prior to division.

We next examined whether overexpression of phyl is sufficient to reduce Ac protein levels in SOPs. In phyl-misexpression clones (see Materials and methods), Ac protein levels in SOPs were suppressed to weaker or even undetectable levels at 12-13 hours APF(Fig. 3F,F′, GFP-positive region), whereas Ac remained at normally high levels in the neighboring wild-type SOPs (GFP-negative region). The lower levels of Ac protein in phyl-misexpressed cells were not due to failure in the specification of SOPs as the density of Sens-positive cells in the phyl-misexpression region were comparable with that in the neighboring tissues without phyl misexpression(Fig. 3F). Therefore, increased phyl expression can lead to more efficient downregulation of Ac protein levels in SOPs, implying that Phyl is the rate-limiting factor in Ac downregulation in SOPs.

Our results have shown that proneural proteins are depleted from SOPs before cell division through proteasome-dependent process, and phylis necessary and sufficient for this degradation. As phyl functions with sina to promote degradation of Ttk, we next asked whether sina was also involved in phyl-mediated Ac downregulation. To answer this question, we focused our analyses on the AWM where Ac proteins are expressed equally in the dorsal and ventral rows of proneural clusters(Fig. 5A). Ac and Sc protein levels in AWM SOPs were also increased in phyl² mutant SOPs (data not shown). By using the ap-Gal4 driver that is expressed in the dorsal compartment of the wing disk (GFP-positive area in Fig. 5A), misexpression of phyl suppressed the Ac clusters in the dorsal row, while the ventral row maintained high levels of Ac (Fig. 5B). In the viable sina²/sina³null-mutant wing disks, misexpression of phyl by ap-Gal4,however, failed to downregulate Ac levels in the SOPs located in the dorsal row (Fig. 5C), suggesting that phyl-mediated Ac downregulation in SOPs requires sina.

We then examined the Ac protein levels and the timing of SOP division in sina mutant clones. In the wild-type thoracic tissues at 16-18 hours APF (Fig. 5D, GFP-positive area), Ac protein had disappeared from the divided SOPs as shown by the Sens-positive cell clusters. However in sina³ mutant clones (GFP-negative area), Ac protein accumulated in single Sens-positive SOPs. Hence, sina is also required for Ac protein downregulation and timely mitotic entry in SOPs.

Phyl acts as a substrate adaptor for the E3 ligase Sina to promote protein degradation (Li et al., 1997; Tang et al., 1997; Li et al., 2002). The result that phyl promotes Ac downregulation in a sina-dependent manner prompted us to ask whether Phyl also acts as adaptor between proneural proteins and Sina. To answer this question, we first tested whether Phyl interacts with the proneural protein Ac. When the myc-ac transgene was co-transfected with flag-phyl into Drosophila S2 culture cells, Myc-Ac protein was co-immunoprecipitated with Flag-Phyl by the anti-Flag antibody (Fig. 6A,lane 2), indicating an association between Phyl and Ac in Drosophilacells. Association between Phyl and Sc was also observed(Fig. 6B, lane 2).

Ac and Sc form heterodimer with the ubiquitiously expressed bHLH protein Daughterless (Da) to activate transcription of downstream target genes(Caudy et al., 1988; Cabrera and Alonso, 1991). We were interested to test whether Phyl also interacts with Da. When flag-phyl was co-transfected with hemagglutinin-da(ha-da) into S2 cells, HA-Da failed to be co-immunoprecipitated with Flag-Phyl (Fig. 6C, lane 2),whereas Myc-Ac was still efficiently co-immunoprecipitated by anti-Flag antibody in a parallel experiment (Fig. 6C, lane 3). Therefore, our results suggest that Phyl does not physically interact with Da. Da protein is ubiquitously expressed in wing disk, and its expression levels in AWM dorsal and ventral proneural clusters and SOPs were slightly higher than that in the disk epithelial cells(Fig. 4D,E). Unlike the Ac and Sc proteins that were depleted from mature SOPs at 2-4 hours APF (data not shown), Da protein was clearly detected in both young and mature SOPs(Fig. 4D,E), suggesting that Da levels are not downregulated in mature SOPs.

To test whether Phyl directly interacts with Ac and Sc, we performed the GST pull-down assay. The bacterially expressed GST-Ac and GST-Sc strongly interacted with Phyl but not the negative control Luciferase (Luc)(Fig. 6D, lanes 7, 8, 10, 11). As a control, a fivefold excess of GST failed to pull down Phyl(Fig. 6D, lane 5). In addition,GST-Ac and GST-Sc proteins also pulled down Sina(Fig. 6D, lanes 9, 12),suggesting that Ac and Sc proteins could form a ternary complex with Phyl and Sina in vitro. By contrast, Da failed to interact with GST-Sina in the pull-down assay (Fig. 6E, lane 7).

To further investigate the role of Phyl in the formation of Phyl/Sina/Ac ternary complex in the context of a whole cell, we analyzed protein interaction using a yeast bridge assay, in which the interaction between the Gal4 DNA-binding domain-Sina (GDB-Sina) fusion protein and the Gal4 activation domain-Ac (GAD-Ac) fusion protein was assessed in the absence or presence of Phyl. In this assay, the expression of phyl was conditionally induced from the MET25 promoter in the absence of 1 mM methionine (-Met) in the yeast plates. As assayed by the growth on the -His -Ade or -His -Ade -Met selection plates, very weak interaction with a few colonies was detected between GDB-Ac and GAD-Sina in the absence of phyl(Fig. 6F, arrows in samples 1 and 3 in plate II and sample 3 in plate III), suggesting that the interaction between Sina and Ac is transient in yeast cells. However, strong interaction between GDB-Ac and GAD-Sina was observed when Phyl was present(Fig. 6F, sample 1 in plate III). No interaction was observed in the control yeasts in which acor sina was not present (Fig. 6F, samples 2 and 4 in plates II and III). Thus, we conclude that Phyl acts as an adaptor to bring Ac and Sina together to form a stable ternary complex.

So far, our results are consistent with the hypothesis that phylmediates timely G2-M transition by promoting proneural protein degradation in SOPs. If higher levels of Ac and Sc in phyl-mutant SOPs are responsible for the delay of SOP division, reduction in ac and sc gene doses should suppress this defect. To test this, we first examined the percentage of divided SOPs in phyl mutants by co-staining the thoraces with antibodies for Sens and Hindsight (Hnt)(Pickup et al., 2002) that label all SOP daughter cells. At 24-26 hours APF when all wild-type SOPs had divided into small clusters composed of 3 to 5 cells, only 27±13% SOPs in female phyl²/phyl⁴ pupae had divided (Fig. 7A,B; column 1 in Fig. 7G). Introducing one allele of sc^10-1 that lacks both ac and sc activities significantly suppressed the division defect of phyl²/phyl⁴, with the percentage of divided SOPs reaching 62±20% (Fig. 7C, and column 2 in Fig. 7G). Some SOPs had even divided into three- or four-cell clusters(arrows in Fig. 7C), a phenotype seen in wild type but not in phyl²/phyl⁴ thoraces at 24-26 hours APF, suggesting that SOP division in these clusters occurs several hours earlier than that in phyl²/phyl⁴mutants. Although the division percentage was increased, the number of Sens and Hnt-positive cells in sc^10-1/+; phyl²/phyl⁴ thorax was dramatically lower than that in phyl²/phyl⁴(compare Fig. 7C with 7B), indicating that the reduction of ac and sc gene doses enhances defects in SOP specification, and SOP specification and division is decoupled in sc^10-1/+; phyl²/phyl⁴ mutants. Taken together,our results showed that decrease of ac and sc gene doses in phyl mutants specifically rescues the G2-M transition delay caused by reduced phyl activity.

Although ac and sc appear to act redundantly to specify SOPs, detailed analyses of ac and sc single mutant found that ac is dispensable for SOP specification when sc is intact (Marcellini et al.,2005). To examine further the roles of ac or scin suppressing G2-M transition, single-mutant allele was introduced into male phyl²/phyl⁴ mutants. We analyzed the male pupae at 22-24 hours APF when the division percentage of male phyl²/phyl⁴ mutants (28±7%) was comparable with that in female mutants at 24-26 hours APF(Fig. 7D; column 3 in Fig. 7G). In phyl²/phyl⁴ male pupae hemizygous for the sc^M6 null allele that affects only scexpression (Gomez-Skarmeta et al.,1995), the percentage of divided SOPs was increased to 43±12% (Fig. 7F, and column 5 in Fig. 7G). Although milder than sc null allele, SOP division defect was also suppressed in phyl²/phyl⁴ male pupae hemizygous for the ac^cami null allele(Marcellini et al., 2005)(37±12%, Fig. 7E and column 4 in Fig. 7G). In addition, clusters composed of three or four SOP daughter cells were also observed (arrows in Fig. 7E and 7F), suggesting that SOPs in these clusters divide a few hours earlier. Therefore, eliminating either ac or sc rescues SOP division defect in phylmutants, suggesting that both genes contribute to inhibition of G2-M transition in SOPs.

Numerous studies in both vertebrate and invertebrate systems have shown that proneural protein-mediated cell cycle arrest is essential for neural precursor specification and differentiation(Ohnuma et al., 2001). However, after the specification process is complete, at least in the case of Drosophila ES organ development, neural precursor has to enter mitosis in order to generate daughter cells that constitute the sensory organs. In this study, we found that cell cycle progression of neural precursor is de-repressed by degradation of proneural proteins. This degradation process requires the E3 ubiquitin ligase complex that includes the RING protein Sina and the adaptor Phyl. In phyl mutant SOPs, Ac and Sc accumulate and G2-M transition is absent or delayed (see Fig. 8 for the summary of the phenotypes). Phyl is a direct transcription target of proneural protein(Pi et al., 2004). Together,our results support a model in which the timing of G2-M transition of SOPs is determined through a negative-feedback loop initiated by proneural proteins.

In this study, several lines of evidence have suggested that prior to the SOP mitosis, Ac and Sc protein levels are efficiently downregulated by the proteasomal degradation pathway. First, ubiquitously expressed Myc-Ac and Myc-Sc proteins were depleted in mature SOPs (Figs 3, 4), suggesting that the downregulation mechanism most probably occurs at the protein level. Second, Ac protein highly accumulated in SOPs when proteasome activity was disrupted(Fig. 3). Third, Ac and Sc proteins accumulated in phyl or sina mutant SOPs (Figs 3, 4 and 5), and Ac and Sc proteins interacted with Phyl and Sina (Fig. 6). The interaction between the Phyl/Sina complex and the Ac and Sc proteins are specific as neither Phyl nor Sina interacts with bHLH protein Da (Fig. 6). Consistent with the protein interaction data, Da protein levels were not downregulated in mature SOPs (Fig. 4). Together,these results lead us to propose that timely degradation of proneural proteins by 26S proteasome is important for cell cycle regulation. Proteolysis also plays important roles in regulating stability of Ac and Sc homologs. The mammalian Ac and Sc homolog Mash1 is degraded in response to BMP signaling during formation of olfactory receptor neurons(Fishell, 1999; Shou et al., 1999), and in neuroendocrine lung carcinoma cells(Vinals et al., 2004). Degradation of human achaete-scute homolog 1 (hASH1) in response to Notch signaling is also observed in small-cell lung cancer cells(Sriuranpong et al., 2002). However, it is still not clear which E3 ligase(s) is responsible for Mash1 and hASH1 degradation.

Although sina plays an important role in phyl-mediated Ac downregulation, several observations suggest that Phyl may also have Sina-independent function. First, the penetrance of the SOP division phenotype was weaker in sina mutants when compared with that in phylmutants. All the remaining SOPs in sina²/sina³ null-mutant thorax divided at 18-20 hours APF, whereas about 50% phyl² null mutant SOPs remained undivided at 24-28 hours APF (Figs 1, 5). Second, although Ac levels at AWM were maintained in phyl mutants at 2-4 hours APF, the protein had disappeared from sina-mutant SOPs at the same stage (data not shown). Third, in fate specification of ES organ SOPs and photoreceptors R1,R6 and R7, phyl loss-of-function mutants display high phenotypic penetrance (Li et al., 1997; Tang et al., 1997; Pi et al., 2001). However, in sina-null mutants, only R7, but not R1 and R6, and only a subset of SOPs are affected (Carthew and Rubin,1990; Pi et al.,2001). These results suggest that Phyl may have Sina-independent functions in these processes. However, these sina-independent functions are not mediated by sina-homolog (sinaH)(Cooper, 2007). Based on the observations that Ac and Ttk proteins accumulate much less in sinamutants than in phyl mutants (data not shown)(Li et al., 1997), it is likely that Phyl may collaborate with other E3s in regulating Drosophila neural development.

Studies in vertebrates have shown that proneural proteins direct permanent cell cycle arrest at G1 phase by elevating the levels of p27Kip1, a CDK inhibitor of Cip/Kip family (Farah et al.,2000). Our study in ES organ development showed that proneural proteins Ac and Sc play dual roles in G2-M transition of SOPs: they initially arrest SOPs at the G2 phase during fate specification; at a later stage, they activate phyl transcription to mediate their own degradation and allow SOP to enter mitosis. In this study, we found that when endogenous Ac and Sc levels accumulated in phyl mutant SOPs, stg levels were low and cells were arrested at G2(Fig. 2). Forced expression of stg as well as reduction of ac and sc gene doses rescued this SOP division defect (Figs 2, 7). Therefore, our data strongly support the model that Phyl upregulates stg expression through downregulation of Ac and Sc protein levels.

We previously have shown that CycE levels in SOPs are greatly reduced in phyl mutants (Pi et al.,2004). However, misexpression of cycE in phyl²/phyl⁴ mutant pupas did not increase the percentage of SOP division (data not shown), suggesting that the SOP division defect seen in phyl mutants is not resulted from reduced levels of CycE. In fact, CycE in SOPs might be involved in regulation of cell cycles of pIIa and pIIb, the daughter cells of SOPs(Audibert et al., 2005). It remained to be determined whether cell cycle progression in pIIa and pIIb cells is altered in phyl mutants.

In the ac and sc mutants, forced expression of phyl is capable of activating asense (ase) [and/or lethal of scute (l'sc)] to induce ES organ formation(Pi et al., 2004). ase is a target gene of Ac and Sc in SOP differentiation, and encodes bHLH protein that shares 70% similarity with Ac in the bHLH domain. We speculate that, in the absence of ac and sc, Ase (and/or L'sc) can substitute for Ac and Sc to regulate G2-M transition of SOPs.

In phyl mutants, SOP-specific gene expression is largely missing(Pi et al., 2001; Pi et al., 2004) and G2-M transition was delayed or absent, indicating that Phyl links specification and mitotic entry of SOPs. The activity of Phyl to mediate both processes seems to depend on its ability to promote protein proteolysis. In SOP specification,the degradation of Ttk is crucial. Our results that reduction of acand sc gene doses in phyl mutants specifically rescued the defects in G2-M transition, illustrate that the degradation of Ac and Sc proteins is necessary for SOP division. Furthermore, when Ac and Sc gene doses were reduced by half, SOP specification and division was effectively uncoupled in phyl hypomorphic mutants (Fig. 7), demonstrating the key role of Phyl in coordinating these two processes. Although phyl is required for degradation of both Ttk and proneural proteins, these two proteins are sequentially degraded in SOPs: Ac and Sc are still enriched in SOPs when Ttk has been depleted(Badenhorst et al., 2002). It implies that the mechanism of Phyl-mediated protein degradation depends on the substrate. Sequential degradation of distinct substrates by the same E3 ligase complex, through mechanisms such as protein modification or co-factor interaction, is also commonly observed in cell cycle progression(Nakayama, 2001; Peters, 2002). As premature entry into mitosis disrupts SOP specification(Negre et al., 2003), the phyl-mediated stepwise degradation of Ttk and proneural proteins provides the precise temporal regulation of cell division during SOP development.

We thank Cheng-Ting Chien, Bertrand Chin-Ming Tan, Henry Sun and Chen-Ming Fan for critical comments on this manuscript. We also thank Pat Simpson for flies and Hugo Bellen for discussion and reagents. We are grateful for I-Chun Chen and all other members of Pi's laboratory for advices and technical support. This study is supported by grants form National Science Council of Taiwan, Chang-Gung University and Chang-Gung Memorial Hospitals.