Drosophila larval brain neuroblasts divide asymmetrically to balance between self‐renewal and differentiation. Here, we demonstrate that the SCFSlimb E3 ubiquitin ligase complex, which is composed of Cul1, SkpA, Roc1a and the F‐box protein Supernumerary limbs (Slimb), inhibits ectopic neuroblast formation and regulates asymmetric division of neuroblasts. Hyperactivation of Akt leads to similar neuroblast overgrowth and defects in asymmetric division. Slimb associates with Akt in a protein complex, and SCFSlimb acts through SAK and Akt to inhibit neuroblast overgrowth. Moreover, Beta‐transducin repeat containing, the human ortholog of Slimb, is frequently deleted in highly aggressive gliomas, suggesting a conserved tumor suppressor‐like function.
This report provides evidence that loss of the SCFSlimb E3 ubiquitin ligase complex as well as hyperactivation of Akt lead to neuroblast overgrowth and defects in asymmetric cell division. Slimb, the F‐box protein of the SCF complex, associates with Akt in a protein complex, and SCFSlimb acts through SAK and Akt to inhibit neuroblast overgrowth.
The SCFSlimb E3 ubiquitin ligase inhibits neuroblast (NB) overgrowth and regulates asymmetric division of NBs.
The NB overgrowth phenotype in the absence of SCFSlimb is similar to what is seen in conditions of AKT hyperactivation.
The F‐box protein Slimb associates with AKT and regulates NB overgrowth via SAK and AKT.
The neural stem cells, or neuroblasts (NBs), of the Drosophila larval brain have emerged as an important model for studying stem cell self‐renewal and tumorigenesis , , , . There are at least two NB lineages with distinct spatial positions and intrinsic properties in the Drosophila larval central brain , , . In both lineage types, NB overproliferation can be triggered by perturbation of asymmetric divisions , . Asymmetric division ensures the polarized distribution of “proliferation factors,” including atypical protein kinase C (aPKC), and “differentiation factors,” including basal proteins Numb, Miranda (Mira), Brain tumor (Brat), and Prospero (Pros), to the daughter NB and GMCs, respectively . The failure of asymmetric division in NBs can result in their hyperproliferation and in the induction of brain tumors . Correct asymmetric protein segregation also relies on mitotic spindle orientation. Inscuteable (Insc), the heterotrimeric G proteins Gαi and Gβγ, and their regulators, Partner of Insc (Pins), Ric‐8, and Mushroom body defect control mitotic spindle orientation in NBs . Several centrosomal proteins, including Aurora A, Polo, Anastral Spindle 2, also regulate spindle orientation and NB self‐renewal , , , , .
The ubiquitination step is mediated by a ubiquitin‐activating enzyme (E1), a ubiquitin‐conjugating enzyme (E2) and a ubiquitin ligase (E3). However, the functions of these enzymes during NB self‐renewal and differentiation were almost completely unknown. Here, we have demonstrated that all three enzymes suppress NB overproliferation and regulate the asymmetric division of NBs.
Results and Discussion
The SCFSlimb E3 ligase complex suppresses NB overproliferation in larval brains
Knockdown (KD) of cullin1 (cul1) using the NB driver insc‐Gal4 resulted in a prominent NB overproliferation in larval brains (supplementary Fig S1B). Approximately 300 NBs that expressed the NB marker Deadpan (Dpn) were observed in a pupal lethal cul1EY11668 mutant at 96 h after larval hatching (ALH) (supplementary Fig S1C; n = 20) compared with approximately 100 NBs in wild‐type (WT) central brains (supplementary Fig S1A; n = 20). Animals that were trans‐heterozygous for cul1EY11668 and a null allele cul1Ex (henceforth referred to as cul1−) displayed 473 ± 100 NBs (Fig 1B and E; n = 23), which was fully rescued by a Flag‐cul1 transgene (supplementary Fig S1D–F). In a time course experiment, cul1− exhibited a dramatic increase in NB numbers at late larval stages (Fig 1E). The numbers of cells labeled by phospho‐Histone H3 (pH3), dMyc, and EdU were significantly increased in cul1−, compared with WT brains (Fig 1C–D; supplementary Fig S1G–J). A dramatic decrease in the number of neurons labeled by Embryonic lethal abnormal vision (Elav) or nuclear Pros was observed in cul1− brains (supplementary Fig S1K–N). There was only one NB in each WT type I (Fig 1F; 100%, n = 32) or type II (Fig 1J; 100%, n = 19) NB MARCM (Mosaic Analysis with Repressible Cell Marker ) clone. In contrast, cul1Ex clones contained ectopic NBs in both type I (Fig 1G; 40%, n = 20) and type II (Fig 1K; 44%, n=18) lineages. cul1 KD using a “type II driver” (see Materials and Methods) also resulted in the generation of multiple NBs (supplementary Fig S3B; 54.7%, n = 75).
We next assessed the function of two other Drosophila SCF subunits, SkpA and Roc1a. Clones derived from skpA1, a loss‐of‐function allele, displayed ectopic NBs in both type I (Fig 1H; 50%, n = 16) and type II lineages (Fig 1L; 62%, n = 21). SkpA KD also resulted in 450 ± 83 NBs (supplementary Fig S3H; n = 20), in contrast to 117 ± 15 NBs in the control (supplementary Fig S3G; n = 20). skpA KD under the type II driver also resulted in NB overproliferation (supplementary Fig S3C; 33.3%, n = 66). A different skpA RNAi line resulted in NB underproliferation in a RNAi screen , likely due to the effect of unknown positional insertion. Likewise, roc1aG1 clones possessed extra NBs in both type I (Fig 1I; 56%, n = 18) and type II (Fig 1M; 64%, n = 22) NB clones. roc1a KD under the type II driver also caused the formation of ectopic NBs (supplementary Fig S3D; 56%, n = 33).
F‐box proteins confer substrate specificity to various SCF E3 ubiquitin ligases . Among 35 F‐box proteins we identified Supernumerary limbs (Slimb) through RNAi targeting and mutant analyses (supplementary Table S1). Clones of the loss‐of‐function allele slimb00295 exhibited a NB overgrowth in both type I (Fig 1N; 42%, n = 24) and type II (Fig 1O; 30%, n = 23) lineages. In the stronger allele slimb8, ectopic NBs were also observed in both type I (supplementary Fig S2B, B’; 45%, n = 20) and type II lineages (supplementary Fig S2D, D’; 50%, n = 24). Likewise, slimb KD under insc‐Gal4 resulted in 271 ± 21 NBs per brain hemisphere compared with control brains (101 ± 8 NBs; supplementary Fig S2E–G). slimb KD under the type II driver also generated supernumerary NBs (supplementary Fig S3F). Loss of three other F‐box proteins, Skp2, Nutcracker (Ntc) or Ago, did not result in NB overgrowth (supplementary Fig S3I–O).
The SCFSlimb complex regulates asymmetric division of NBs
In cul1Ex clones, 55% of metaphase NBs displayed either uniformly cortical or diffused cytoplasmic aPKC localization (Fig 2B; n = 31), in contrast to WT NBs (Fig 2A; 100%, n = 18). Consistently, 39% of cul1Ex metaphase NBs exhibited delocalized Numb (Fig 2C; n = 31), in contrast to control NBs (Fig 2A; 100%, n = 18). The localization of other polarized proteins, including Bazooka (Baz), Par6, Insc, and Pins, was mildly disrupted as cortical distribution or misoriented crescents in the cul1− NBs (supplementary Fig S4A–H). cul1− NBs also displayed a spindle misorientation at metaphase (Fig 2J, J’; 40%, n = 95; WT, Fig 2I, I’) with 13% of NBs showed a 90° misalignment (orthogonal division) of the mitotic spindle with the apicobasal axis (Fig 2J, J’; n = 95). This defect in cul1 mutants was fully rescued by the expression of Flag‐tagged Cul1 (Fig 2K, K’; n = 31). aPKC (5.4%, n = 205) and Numb (5.4%, n = 205) were missegregated into both daughter cells during telophase in cul1− NBs (Fig 2H) compared with the control (Fig 2G; n = 136).
In time‐lapse experiments on living whole‐mount brain explants expressing α‐tubulin‐GFP, control NBs always divide asymmetrically (Fig 2L; n = 21). In contrast, 22% of cul1− NBs divided to generate two similar‐sized daughters (Fig 2M; n = 23), a remarkable phenotype in asymmetric division. Although cul1− NBs showed a delay in mitosis, cell division defects are unlikely a major cause of NB polarity defects in mutants for the SCF complex, because several known cell division mutants did not affect cell polarity , , .
roc1aG1 clones also exhibited delocalization of aPKC (Fig 2D; 28.6%, n = 14) and Numb (Fig 2D; 28.6%, n = 14) at metaphase. Similarly, in skpA1 clones, 37.5 and 12.5% of metaphase NBs displayed delocalized aPKC and Numb, respectively (Fig 2E; n = 16). Interestingly, aPKC (Fig 2F; 71%, n = 7) and Numb (Fig 2F; 57%, n = 7) were dramatically delocalized in metaphase NBs derived from slimb8 clones. Spindle misorientation defects were also observed in skpA1 (supplementary Fig S4J; 43%, n = 14) and slimb00295 (supplementary Fig S4K; 29%, n = 28) clones. Therefore, we conclude that the SCFSlimb complex plays an important role in regulating NB asymmetric division.
In aPKC06403 cul1− double mutant, the NB number was significantly reduced at 70°h ALH (supplementary Fig S5C and D; 119 ± 42, n = 22), compared with the cul1− control (supplementary Fig S5B and D; 200 ± 20, n = 21). Further, the NB overgrowth in cul1− was largely suppressed by notchts1 at 84 h ALH at 29°C (supplementary Fig S5E–H). These data suggest that Cul1 may function upstream of aPKC and Notch signaling or redundantly with them to inhibit NB overgrowth.
Uba1, Eff and Nedd8 suppress NB overproliferation and regulate asymmetric division
Uba1 is the only E1 enzyme in Drosophila and in a loss‐of‐function uba1s3484 allele, ectopic NBs were observed in both type I (Fig 3B; 40%, n = 42) and type II clones (Fig 3F; 26%, n = 31). Among 16 genes encoding E2 enzymes (supplementary Table S2), RNAi targeting of two of them, effete (eff; also called ubcD1) (supplementary Fig S6B, B’) and ubcD10 (data not shown), resulted in type II NB overgrowth. Ectopic NBs were observed in both type I (Fig 3C; 33%, n = 36) and type II (Fig 3G; 40%, n = 32) effD73 clones, but not in clones of ubcD10BG00902 (supplementary Fig S6E–F). Given that the ubcD10 KD construct has predicted off‐target sites, the phenotype caused by ubcD10 KD was unlikely specific. Clones for two other E2 mutants, ubcD2 and ubc9, did not obviously change NB numbers (supplementary Table S2 and data not shown). Collectively, Eff is a specific E2 enzyme that regulates NB self‐renewal cell‐autonomously. Neddylation, a process of Nedd8 conjugation at a conserved lysine residue of Cullins, is essential for Cullin‐based E3 ligase activities . Ectopic NBs were observed in both type I (Fig 3D; 29%, n = 28) and type II clones (Fig 3H; 40%, n = 25) that were derived from a nedd8 hypomorphic allele, nedd8AN015.
In uba1s3484 metaphase NBs, aPKC was delocalized throughout the cell cortex and became cytoplasmic (Fig 3J; 43%, n = 35), and Numb was partially delocalized (Fig 3J; 29%, n = 35). Similarly, in effD73 NBs, aPKC (Fig 3K; 59%, n = 29) and Numb (Fig 3K; 21%, n = 29) were no longer asymmetrically localized at metaphase. In nedd8AN105 clones, both aPKC (Fig 3L; 42%, n = 24) and Numb (Fig 3L; 25%, n = 24) proteins were often delocalized. Furthermore, spindle misorientation was observed in both uba1s3484 (Fig 3N) and nedd8AN015 (Fig 3O) NBs at metaphase. Interestingly, NBs from either uba1s3484 or effD73 clones formed multiple centrosomes (supplementary Fig S6H–I), similar to those from slimb mutants , . Therefore, Uba1, Eff and Nedd8, similar to the SCFSlimb E3 ligase, regulate asymmetric division/self‐renewal of NBs as well as centrosome numbers.
SAK is a target of the SCFSlimb complex during NB self‐renewal
The SCFSlimb complex targets the SAK kinase (Polo‐like kinase 4) for degradation during centriole formation , . We therefore assessed whether the SCFSlimb complex controls NB self‐renewal through SAK. SAK‐GFP overexpression led to spindle misorientation in NBs (supplementary Fig S7B, B’; 24%, n = 63; ), but not NB overproliferation, likely due to the small proportion of cells undergoing orthogonal divisions (supplementary Fig S7B, B’; 5%, n = 63). Neither did it cause any apparent delocalization of aPKC, Numb or Mira in metaphase NBs (supplementary Fig S7O–P and data not shown). SAK overexpression under insc‐Gal4 with concomitant cul1 KD in larval brains resulted in a significant increase in the number of NBs compared with the control (supplementary Fig S7C, D and G), suggesting a genetic enhancement. Similarly, NB overgrowth in skpA KD was exacerbated by overexpressing SAK (supplementary Fig S7E–G). Moreover, defects of NB overgrowth (supplementary Fig S3P–T) and multiple centrosomes (supplementary Fig S7H–L) were remarkably suppressed in the cul1EY11668 sakc06612 double homozygotes. Thus, the SCFSlimb complex controls NB self‐renewal through SAK and additional unknown targets.
Akt hyperactivation leads to the formation of supernumerary NBs and asymmetric division defects
Akt is a critical regulator of cell proliferation, growth and metabolism  and is required for NBs to exit from quiescence at early larval stages , . Akt KD was found to give less NBs by a RNAi screen . Overexpression of a myristoylated, constitutively active form of Akt (Myr‐Akt) in central brain NBs, starting from 24 h after egg laying (AEL) using insc‐Gal4 under the control of a tub‐Gal80ts element, resulted in 659 ± 169 NBs at 96 h ALH at 29°C (Fig 4B; n = 21) compared with 91 ± 8 NBs in control brains (Fig 4A; n = 20). Upon Akt hyperactivation, the number of cells labeled by pH3, CycE and EdU were significantly increased while the number of neurons labeled by Elav or Pros was strongly decreased compared with the control (Fig 4C–L). Myr‐Akt overexpression under the type I‐specific driver ase‐Gal4 (supplementary Fig S8B, B’; 31%, n = 75) or the type II driver (supplementary Fig S8D, D’; 55%, n = 53) resulted in NB overgrowth.
In Myr‐Akt NBs, aPKC (Fig 4N; 55%, n = 51), Numb (Fig 4P; 45%, n = 51), Mira (Fig 4R; 50%, n = 32) and Pon (Fig 4T; 29%, n = 45) were delocalized at metaphase. Furthermore, Myr‐Akt expression resulted in mitotic spindle misorientation in 36% of NBs with 9% orthogonal divisions (Fig 4V, V’; n = 53). Thus, Akt hyperactivation significantly disturbs asymmetric protein localization and mitotic spindle orientation during NB division. However, hyperactivation of Akt does not affect centrosome number in neuroblasts (supplementary Fig S8F), suggesting that Akt and SAK likely affect spindle orientation through different mechanisms.
The SCFSlimb complex inhibits ectopic NB formation, in part through Akt
We next assessed whether Akt functions downstream of the SCFSlimb complex. In cul1EY11668; akt1/akt3 double mutant, the NB number was restored to a number close to that of the WT control (Fig 5E and G; 100 ± 15; n = 20), in contrast with either the cul1EY11668 (Fig 5B and G; 307 ± 21; n = 20) or akt1/3 (Fig 5D and G; 38 ± 5; n = 20) single mutants. Partial suppression was even observed via the heterozygous akt1 mutation (Fig 5C, F and G; 186 ± 25.6, n = 20). Knocking down of akt by RNAi also dramatically suppressed neuroblast overproliferation in cul1 RNAi but not brat RNAi background (supplementary Fig S5I–L). Moreover, using an antibody against phosphorylated Akt at Serine 505, a significant increase in the phosphorylated Akt signal was observed in cul1 brains compared with the control (Fig 5H–J). Consistently, p505‐Akt signal was dramatically increased in both slimb RNAi brains (Fig 5G–H) and slimb8 mutant clones (Fig 5K–L’), but decreased in both akt3 clones and akt KD (supplementary Fig S8I–L).
We showed recently that Akt associates with Slimb and can be ubiquitinated by Slimb (Fig 5L; ). Another human E3 ligase tetratricopeptide repeat domain 3 facilitates the ubiquitination and degradation of Akt . Next, we determined which domain of Akt is important for its association with Slimb. The central region of Akt, which contains its protein kinase domain (T2), but not the N‐terminus (T1, containing a pleckstrin homology domain) or the C‐terminus (T3, containing an AGC kinase domain), interacted with Slimb in S2 cells by co‐immunoprecipitation (Fig 5K–L). Taken together, our biochemical and genetic data suggest that the SCFSlimb complex inhibits ectopic NB formation in part through an association with Akt (Fig 5M).
Conserved function of human BTRC in gliomas
Beta‐transducin repeat containing (BTRC/β‐TrCP; human homologue of Slimb) showed a significant loss in 72.5% of glioma patients (supplementary Fig S9A; n = 261) in the glioma patient database REMBRANDT . Its copy number was an independent predictor of prognosis in a multivariate analysis (P = 0.0434). BTRC copy loss was observed in patients with glioblastoma (82%) and oligodendroglioma (68%). It was also observed in patients with mesenchymal and proliferative (85 and 88%, respectively) (Table 1), which are frequently associated with activated AKT signaling, a central oncogenic pathway regulating glioblastoma (GBM) growth and survival . The BTRC functional module was derived by mining mRNA expression databases and mapped to 544 mRNA transcripts (supplementary Table S3). This module was able to stratify patients into two subgroups and was significantly associated with survival (supplementary Fig S9B). A multivariate Cox Regression model confirmed that the BTRC functional subgroup was independently associated with survival (supplementary Table S4). Furthermore, BTRC copy changes have a significant inverse correlation with the gene expression of AKT2 and PIK3CD (supplementary Fig S9C).
β‐TRCP, the mammalian Slimb homolog, is speculated to play a greater role as an oncogene than as a tumor suppressor . Here we demonstrate that the SCFSlimb complex plays an important role in NB self‐renewal and asymmetric division. It acts at least in part through the oncogenic protein Akt, a key player that regulates NB self‐renewal and asymmetric division. The activation of Akt in NBs appears to be independent of Phosphatase and tensin homologue (PTEN), a negative regulator of PI3K/Akt signaling, because no NB overgrowth were observed in two loss‐of‐function alleles, PTENC494 and PTEN1 (data not shown). A recent report suggested a role for Target of rapamycin (TOR) signaling in restraining tumorigenesis in larval brains . PI3K/TOR also interacted genetically with Pins in the suppression of tumor growth in larval brains , suggesting the complex role of PI3K/Akt and the interconnected TOR pathway in NB homeostasis.
Materials and Methods
MARCM clones were generated as previously described . Briefly, larvae were heat‐shocked at 37°C for 90 mins at 24 h ALH and at 10–16 h after the first heat shock. Larvae were further aged for 3 days at 25°C, and larval brains were dissected and processed for immunohistochemistry. For gene KD under the type II diver (w; UAS‐Dicer 2, wor‐Gal4, ase‐Gal80/CyO; UAS‐mCD8‐GFP/TM3, Ser ), UAS lines were crossed to type II NB driver at 25°C and shifted to 29°C at 24 h ALH for 3 or 4 days.
Live imaging recording
Live imaging recording was performed as previous described . NBs of the central brain were sampled using a 40× objective on a LSM 700 confocal microscope. Pictures and videos were exported and processed with Abode Photoshop and ImageJ respectively.
HW, CT, BTA and FY designed the experiments and analyzed the data. LS, WC, ES, SA, CTK and JW conducted the experiments. LS, CT and HW wrote the paper.
Supplementary Figure 1
Supplementary Figure 2
Supplementary Figure 3
Supplementary Figure 4
Supplementary Figure 5
Supplementary Figure 6
Supplementary Figure 7
Supplementary Figure 8
Supplementary Figure 9
Supplementary Materials and Methods and Supplementary Tables
We thank J Skeath, H Stocker, I Hariharan, CT Chien; RJ Duronio; B Limbourg‐Bouchon, M Bettencourt‐Dias, A Wodarz; B Edgar, C Doe, F Matsuzaki, J Knoblich, CY Lee, H Richardson, T Lee, H Steller, A Swan, YN Jan, the Hybridoma Bank, the Bloomington Drosophila Stock Center, Vienna Drosophila RNAi Center for reagents and WL Yong for technical assistance. This work is supported by Duke‐NUS Signature Research Program funded by the Agency for Science, Technology and Research (A*STAR), Singapore and Ministry of Health, Singapore, the National Research Foundation Singapore under its NRF fellowship (Award No. NRF2008NRF‐RF001‐117), and the Singapore National Research Foundation under its ‘New Investigator Grant’ and administered by the Singapore Ministry of Health's National Medical Research Council [NMRC/NIG/1059/2011 (CW)].
- Conflict of interest
The authors declare that they have no conflict of interest.
- Received September 9, 2013.
- Revision received November 18, 2013.
- Accepted November 19, 2013.
- © 2014 The Authors