Drosophila Brat and Human Ortholog TRIM3 Maintain Stem Cell Equilibrium and Suppress Brain Tumorigenesis by Attenuating Notch Nuclear Transport

Glioma stem cells (GSC) are a driving force underlying the clinically devastating growth properties of glioblastoma (1). Although GSCs account for only a small fraction of total glioblastoma cells, they confer a disproportionate influence on malignant behavior and therapy resistance (2–4). As opposed to normal stem cells, GSCs show disrupted asymmetric cell division properties that favor the creation of stem cells over those that are programmed for differentiation (5). Current understanding of asymmetric cell division as it relates to tumorigenesis has been derived largely from studies of Drosophila neuroblasts. In this model, mutation of a single gene, Brain Tumor (brat), causes altered stem cell division in a manner that generates a massively enlarged brain formed entirely of neoplastic neuroblasts during larval development (6, 7).

Mammalian tripartite motif-containing protein 3 (TRIM3) is a human ortholog of Drosophila Brat (48% homology) expressed solely in brain and is deleted in 25% of glioblastoma samples; other mechanisms are responsible for its reduced gene and protein expression in nearly all glioblastomas (5). We previously demonstrated that restored expression of TRIM3 reduced neurosphere formation, attenuated the GSC population, promoted normal asymmetric cell division, and diminished in vitro and in vivo growth properties of human glioblastomas. Initial studies implicated NOTCH1 as a potential mediator of these effects (5).

Notch signaling is a central node that directs self-renewing proliferation of neural stem cells (8, 9). Notch was first described as oncogenic in T-cell acute lymphoblastic leukemia, in which a specific translocation t(7;9)(q34;q34.3) generates a fusion protein with a truncated, active Notch intracellular domain (NICD; ref.10), Notch is now appreciated as a key protumorigenic signaling protein in numerous cancers, including glioblastoma (11–13, 14).

Activation of Notch signaling is complex and requires receptor activation, endocytosis, followed by γ-secretase–mediated cleavage to generate active NICD and, finally, transport of NICD into the nucleus by Importins to initiate transcription (15–19).

In this study, we investigated mechanisms by which Brat/TRIM3 regulates Notch signaling in brain tumors. We generated a novel Drosophila model using brat-RNAi expressed in neuroblasts, which results in a fatal adult brain tumor phenotype, and demonstrated that active Notch is a primary driver. We extended investigations to human glioblastoma neurospheres and show a similar relationship between TRIM3 and NOTCH1 signaling. Finally, we present data supporting a mechanism in which Brat/TRIM3 suppresses Notch signaling by attenuating its nuclear transport through Importin α and β.

UAS-bratRNAi and insc-GAL4; UAS-tub-cherry flies were obtained from Bloomington Drosophila Stock Center at Indiana University (Bloomington, IN). UAS-MamH fly stock was a generous gift from Dr. Barry Yedvobnick, Emory University, Atlanta, GA (20). All Drosophila stocks were reared at 25°C with standard cornmeal/yeast/agar medium. UAS-bratRNAi and insc-GAL4; UAS-tub-cherry were crossed to generate adult flies with brain tumor phenotype. UAS-MamH flies were brought into the same genetic background to suppress Notch signaling pathway in brat-RNAi tumor flies. Repo–GFP flies from Bloomington Drosophila Stock Center were crossed into brat-RNAi tumor model to check Repo-positive cells. To suppress nuclear transport, we used UAS-ket-RNAi stock of Importin β ortholog Drosophila Ketel in brat-RNAi background (Bloomington Stock Center of Indiana University, Bloomington, IN).

Adult Drosophila brains were dissected and fixed in 4% paraformaldehyde in PBS for 60 to 90 minutes (21). These were then treated with 0.5% Triton X-100 in PBS for 30 minutes and placed in primary antibodies in 1× PBS with 0.5% Triton X-100 and 10% BSA overnight at 4°C. Brains were washed for at least 30 minutes, added to secondary antibody solution for overnight at 4°C, washed for the final time and kept in VECTASHIELD (Vector Laboratories) for 2 days at 4°C, and mounted for microscopy.

The following primary antibodies were used for IHC: NICD (1:100; Cell Signaling Technology), Miranda (1:100) and Deadpan (1:100; a gift from R. Read), phospho-histone H3 (PH3; Abcam), Musashi (1:100; a gift from Dr. Hideyuki Okano, Riken Brain Science Institute, Tokyo, Japan), and Brat (1:100; a gift from Dr. Jürgen Knoblich, Institute of Molecular Biotechnology of the Austrian Academy of Sciences (IMBA), Vienna, Austria). Alexa Fluor 488, 555, and 647 were used as secondary antibodies.

Thirty normal brains and 30 brat-RNAi tumor brains were collected, and nuclear extracts were prepared using Thermo Scientific NE-PER Nuclear Extraction Kit (cat# 78833). Proteins were run on a 4%to 15% gradient gel and visualized using ECL. Primary antibodies used were NICD (Cell Signaling Technology), Cut (1:100; a gift from K. Moberg), histone H3 (1:1,000; Abcam), and actin (1:2,000).

Neurosphere cultures were established from normal human neuroprogenitor cells (NHNP) obtained from Lonza (cat# PT-2599) and patient glioblastoma samples N08-74 and 13113, as described previously, and the experiments were performed using passage numbers 10 to 40 (5). The diagnosis of glioblastoma was established by the senior author, who is a diagnostic neuropathologist (D.J. Brat) in accordance to criteria by the World Health Organization. The cell lines are routinely tested for GFAP, Olig2, Sox2, and CD133 expression by Western blot analysis and immunocytochemistry. The cells were last tested within the last 6 months. The cells were tested and cleared for mycoplasma using Lonza MycoAlert Mycoplasma Detection Kit (cat# LT07-703). Glioblastoma neurospheres and NHNPs were cultured using Neurobasal-A Media (Invitrogen) with added EGF (Stem Cell Technologies), FGF (Stem Cell Technologies), and heparin (Stem Cell Technologies), as described previously (5).

FUW-TRIM3 construct was used in all experiments (5). The TRIM3 cDNA was obtained from pCMV6–AC–TRIM3–GFP TrueORF clone (RG211739; Origene). To clone the TRIM3 cDNA, the gene was amplified from the TrueORF clone and recloned in FUW vector using BamH1 and Hpa1 sites. The following primers were used: BamHI–TRIM3 5'-GGATCCGCCATGGCAAAGAGGGAGGACAGC-3' and HpaI–TRIM3 5'-GTTAACCTACTGGAGGTAGCGATAGGCTTT-3'.

For glioblastoma neurosphere, viral particles were collected and concentrated by ultracentrifugation at 25,000 rpm for 4 hours at 4°C. Viral pellets were resuspended in Neurobasal-A Media to prevent serum effects. Infection of cells and lentiviral-mediated gene expression were confirmed by puromycin selection process and TRIM3 expression by Western blot analysis.

Glioblastoma stem cells were sorted using BD FACSCanto II (BD Biosciences) based on stem cell marker CD133. Analysis of stem cell population in neurospheres was performed using antibodies against CD133 (Miltenyi Biotec) in BD FACSCanto II (BD Biosciences) flow cytometry. Antibody used was CD133-PE conjugated (Miltenyi Biotec).

Antibodies against endogenous Importin β1 (Sigma) and endogenous TRIM3 (Abcam) were used for coimmunoprecipitation and reverse coimmunoprecipitation, respectively. Pierce coimmunoprecipitation was used for the pull-down assay, and Pierce mild elution buffer was used to release the protein complex from the beads. Protein was boiled at 95°C for 10 minutes after adding the nonreducing buffer with 20 μmol/L DTT. Proteins were run in 4% to 15% SDS gel along with appropriate controls.

Western blots were performed using 4% to 15% gradient gels from Bio-Rad. Total protein was isolated using RIPA buffer (Sigma), and nuclear and cytosolic extracts were obtained using the NE-PER buffer (Thermo Scientific). The following antibodies were used to detect proteins: TRIM3 (GeneTex), HES1 (5), NOTCH1, NICD, KPNA4 (Importin α3; Everest Scientific; ref.18), KPNB1 (Importin β1; Sigma; ref.16), tubulin (GeneTex), GAPDH (GeneTex), histone H3. Unless mentioned, antibodies were obtained from Cell Signaling Technology.

Cells were isolated from media and a single-cell suspension was created using Accutase. Cells were then fixed with 4% paraformaldehyde, cytospun on slides, permeabilized using 1× PBS with 0.3% Triton X-100, and subjected to blocking at 0.1% BSA in 1× PBS for an hour. Primary antibody incubation was done overnight at 4°C, followed by washing. After secondary antibody incubation for an hour at room temperature, cells were washed and mounted with VECTASHIELD with DAPI. Primary antibodies included NICD (Cell Signaling Technology), KPNB1 (Importin β1; Sigma; ref.16), and CD133 (Miltenyi Biotec). Alexa Flour 488, 555, and 647 were used as secondary antibodies.

GraphPad Prism and FloJo software were used for statistical analysis. For densitometric analysis, ImageJ was used to quantify the Western blot bands and to quantify the net fluorescent intensity in the cellular immunofluorescence. Imaris 3D imaging software was used to quantify the brain volume and total tumor cell population in flies. Obtained results were then analyzed for significance using nonparametric t test at GraphPad Prism.

Drosophila adult brain tumor imaging and human patient-derived neurosphere cell imaging were achieved with the assistance of the Integrated Cellular Imaging Core of Emory University (Atlanta, GA). An FV1000 Olympus Confocal Microscope was used to collect pictures.

Brat protein localizes to the basal side of asymmetrically dividing neuroblasts in Drosophila, initiating a differentiation signal, whereas the apical side lacking Brat generates a neuroblast through a self-renewal program (1). brat-null mutants lose the capacity to divide asymmetrically and accumulate proliferative neuroblasts that leads to a brain tumor phenotype that is lethal at the larval and pupal stage (7). To establish an adult brain tumor model in Drosophila that could be used to study the temporal evolution of signaling events and to translate to mammalian systems, we used brat-RNAi specifically targeted to Drosophila neuronal stem cells using the inscuteable-Gal4 driver that is expressed only during neuroblast division (1, 22). We found that suppressing Brat in neuroblasts resulted in an adult Drosophila brain that was three times the size of a normal one (Fig. 1A–C). Tumoral neuroblasts resulting from brat-RNAi were noted in more than 60% of the brain and involved nearly all regions equally (Fig. 1D), expanding it globally (Fig. 1B). To ensure that brat-RNAi resulted in reduced Brat protein, we used a Brat antibody kindly provided by Dr. Jorgen Knoblich. In the third instar larval brain and in the adult brain (day 5), we found that brat-RNAi expression in neuroblasts consistently and significantly reduced Brat protein expression (Supplementary Fig. S1A–S1X). Control flies, which expressed insc-GAL4–driven m-cherry, were completely normal.

Previous studies and our PH3 staining indicated that adult Drosophila brain does not contain dividing cells, whereas neuroblasts induced by brat-RNAi showed a marked increase in proliferation (Fig. 1E and F vs. G and H and Supplementary Fig. S2C–S2F). As opposed to normal adult Drosophila brain, which do not contain neuroblasts and lack expression of the stem cell markers Miranda (Mira) and Deadpan (Dpn), brat-RNAi tumor cells showed consistent Mira (Fig. 1I and J and Supplementary Fig. S2I and S2J) and Dpn upregulation (Fig. 1K and L), indicating a high degree of stemness. One unexpected observation was that some brain cells expressing Miranda did not show a corresponding expression of Tub-cherry (Fig. 1I and J). Only cells that express inscuteable would be expected to also express UAS-tub-cherry transgene, as it is driven by inscuteable-GAL4. However, daughter cells that do not have inscuteable expression may still harbor enough brat-RNAi to suppress Brat and allow the expression of the stem cell marker Mira. To determine whether proliferating stem cells demonstrate a glial lineage, we used the marker Repo-GFP together with the neuroblast marker Mira and found that approximately 6% of proliferating neuroblasts also showed Repo expression, as compared with none in control brains (Fig. 1M–O and Supplementary Fig. S2G and S2H).

We also examined the effects of brat-RNAi–driven brain tumors on function and survival in adult flies and found that this phenotype was associated with poor motor function, as determined by negative geotaxis (climbing) assay, and that the median posteclosion survival was dramatically reduced (14 days in brat-RNAi flies) as compared with controls (45 days; P < 0.0001), indicating that this phenotype is functionally severe and eventually lethal (Supplementary Fig. S2A and S2B; refs. 23–25). These data confirm that brat-RNAi specifically directed to neuroblasts in Drosophila results in a brain tumor phenotype with disrupted asymmetric cell division that could be used to further explore mechanisms related to disrupted stem cell dynamics in brain tumors.

Brain tumor phenotypes developed by brat and Notch pathway mutants are similar, suggesting shared common pathways (26). For example, Numb is a Notch suppressor and numb mutants develop a phenotype similar to brat, although these fail to complement each other (27). We hypothesized that Brat suppresses Notch, given that Numb-independent Notch pathway activation has been recognized (28). To date, no direct link has been established between Notch and Brat in Drosophila (26). In human glioblastoma, the Brat ortholog TRIM3 suppresses NOTCH1 target HES1, suggesting an effect on Notch signaling (5). In the Drosophila brat-RNAi adult brain tumor model, we found that the suppression of Brat in proliferating neuroblasts resulted in increased levels of active Notch (NICD) within these cells (Fig. 2A–D) and that much more NICD localized to the nuclei when Brat was suppressed (Supplementary Fig. S3A–S3C). We next asked whether the increased active NICD in nuclei following Brat suppressing was responsible for initiating a transcriptional program. Nuclear extract from brain tumor neuroblasts showed that Brat suppression led to increased expression of the transcription factor Cut (Fig. 2E and F), a Notch target, indicating active Notch signaling. These results suggest that Brat negatively regulates Notch signaling. To further investigate, we used a mutant form of Mastermind, the transcriptional partner of NICD that is required for a functional Notch transcriptional complex. The truncated form of Mastermind (MamH; ref.20) lacks the NICD-binding domain and prevents Notch signaling. When we expressed MamH in neuroblasts along with brat-RNAi, the brat RNAi phenotype was diminished, with a significant reduction in both tumor cell numbers and tumor volume (Fig. 2G–J). Combined, our results indicate that Brat mediates its effects on neuroblast proliferation and brain tumor formation at least partially through the suppression of Notch signaling.

The increased accumulation of NICD in brat-RNAi tumor cell nuclei, beyond the level of increases in total protein expression, suggested that nuclear transport of NICD might be enhanced. Nuclear transport of NICD in Drosophila and mammalian systems is controlled by the Importin complex, consisting of the partners Importin α and β (29). To determine the significance of nuclear transport in Brat regulation of NICD, we genetically manipulated the Drosophila homolog of Importin β gene (Ketel) in the background of brat-RNAi. We found that RNAi of Ketel reduced the expression of Ketel protein and partially reversed the phenotype of brat-RNAi (Fig. 2K–N). Together, our results indicate that brat-RNAi increases the expression of NICD and also enhances its transport to the nucleus.

To translate our findings from Drosophila, we used neurosphere cultures derived from human glioblastoma explants to study TRIM3, the human ortholog of Brat. TRIM3 is deleted in 25% of glioblastomas and is expressed at low levels in nearly all (5, 30). Previous work has shown that the restoration of TRIM3 attenuates in vitro and in vivo growth properties of glioblastoma (5). First, we investigated TRIM3 expression in NHNP and found strong TRIM3 protein expression (Fig. 3A). In Drosophila neuroblasts, Brat knockdown increased both total active Notch NICD expression and its nuclear accumulation. To determine whether this also occurred in human neuroprogenitors, we suppressed TRIM3 in NHNP using TRIM3 sh-RNA. TRIM3 sh-RNA successfully reduced TRIM3 protein (Fig. 3A), but did not strongly induce total NICD expression, as noted in Drosophila (Supplementary Fig. S4B). However, we found that suppression of TRIM3 substantially increased NICD nuclear accumulation, similar to the effects in Drosophila (Supplementary Fig. S4C and S4D). As the loss of Drosophila Brat induces stemness in Drosophila neuroblasts, we determined whether NHNP stemness and differentiation had any effect on human TRIM3. We FACS sorted NHNP cells using stem cell marker CD133 and found that CD133⁺ cells have low TRIM3 compared with CD133⁻ cells (Fig. 3B). We also differentiated the NHNP cells for 10 days and found that differentiated NHNP cells had more TRIM3 than undifferentiated cells (Fig. 3B). Our data clearly indicated that TRIM3 expression is inversely related to stemness.

We went on to use patient-derived glioblastoma neurosphere cultures (N08-74 and 13113), which lack TRIM3 protein, to study the effects of TRIM3 restoration on Notch signaling and stemness (Fig. 3C). To restore TRIM3 expression, we relied on stable transduction process that faithfully expressed the protein (Fig. 3C). We also investigated the cellular localization of TRIM3 and demonstrated that it was almost entirely cytosolic (Supplementary Fig. S4E). Using our TRIM3-overexpressing neurosphere lines, we demonstrated that TRIM3 strongly suppressed the expression of the NOTCH1 target HES1 (Fig. 3C), indicating that TRIM3 normally inhibits NOTCH1 signaling in mammalian systems much like that Brat regulates Notch in Drosophila. As human NOTCH1 is a fundamental driver of self-renewal and stem cell maintenance (31, 32), we next reasoned that TRIM3 would reduce the stem cell component in human glioblastomas. Using flow cytometry on N08-74 and 13113 glioblastoma neurospheres, we found that TRIM3 restoration was associated with a marked reduction in the percentage of CD133⁺ glioma stem cells as compared with neurospheres lacking TRIM3 (Fig. 3D–F). These findings indicate that TRIM3 suppresses NOTCH1 signaling and stemness in human gliomas.

Many cancers show the activation of NOTCH1 signaling, similar to our finding of NOTCH1 upregulation in glioblastoma neurospheres lacking TRIM3 (14, 33–36). Interestingly, although NOTCH1 activity was clearly increased, our Western blots for full-length NOTCH1 and cleaved active NOTCH1 (NICD) that used whole-cell (total) protein extracts revealed no substantial differences between their protein levels in TRIM3-expressing N08-74 and 13113 as compared with their TRIM3-null counterparts (Fig. 4A). We hypothesized that NICD is expressed but fails to enter the nucleus to initiate transcription. Using an antibody that recognizes NICD, but not full-length NOTCH1, we performed immunocytochemistry and demonstrated reduced levels of NICD in the nucleus, but not the cytoplasm of N08-74 and 13113 upon TRIM3 expression (Fig. 4B–E1). Western blots of nuclear and cytosolic extracts also demonstrated that TRIM3 expression was associated with reduced expression of NICD in the nucleus (Fig. 4G), but not the cytoplasm (Fig. 4F). These results suggested that TRIM3 might suppress the nuclear transportation of NICD.

Transport of NICD to the nucleus is a critical step for the activation of NOTCH1 signaling. Cleavage of NOTCH1 to generate active NICD occurs in the early endosome and/or in multivesicular bodies (37). After formation, NICD is normally transported into the nucleus by the nuclear transporter Importin complex, composed of Importin α and β. Previous studies indicate that among several types of Importin α and β, NICD nuclear transport depends mostly on Importin α3, α4, α7 and Importin β1 (16, 18). As we observed that NICD localized in the cytoplasm rather than the nucleus after TRIM3 expression, we reasoned that nuclear transportation might be suppressed. To test this, we performed a Western blot analysis for Importin α3 and Importin β1 in glioblastoma neurospheres that differentially express TRIM3. We found that total (nuclear and cytoplasmic) Importin α3 expression was reduced following TRIM3 expression, but that total Importin β1 expression remained stable (Fig. 5A). To further investigate the cytonuclear translocation of Importins, we examined Importin β1 levels in cytosolic and nuclear extract. We found lower Importin β1 inside the nucleus in TRIM3-expressing cells, with only modest to no reduction of cytosolic levels (Fig. 5B–E). We confirmed these results using immunofluorescence, which showed greater nuclear accumulation of Importin β1 in glioblastoma cells lacking TRIM3 (Supplementary Fig. S5A–S5D).

To explore whether TRIM3 directly binds with Importin β1 to retain it in the cytosol, we performed coimmunoprecipitation of endogenous Importin β1. Western blot analysis of immunoprecipitated Importin β1 revealed that it strongly binds with TRIM3 in those neurospheres that express TRIM3, but not those that do not (N08-74 and 13113). In NHNP cells, which normally express TRIM3, binding of TRIM3 and Importin β1 was only modest (Fig. 5F). The reverse coimmunoprecipitation that used a TRIM3 antibody showed that endogenous TRIM3 also binds with Importin β1, providing addition evidence that TRIM3 directly interacts with critical nuclear transporters (Fig. 5G). As Importin α3 was noted in both the Importin β1 and TRIM3 immunoprecipitations, it is likely that TRIM3 binds to the Importin transport complex. An earlier study has shown that Importin β1 is capable of binding with NICD in oligodendrocyte precursor cells (16). In accordance with this, here, we confirmed that in NHNP and in glioblastoma neurospheres, NICD binds with endogenous Importin β1 (Fig. 5F). NICD was also detected in endogenous TRIM3 immunoprecipitation (Fig. 5G), establishing that NICD is a part of the TRIM3, Importin and NICD complex. We also performed immunocytochemistry and found NICD and Importin β1 staining overlapped (Supplementary Fig. S5E–S5H). Together, these data indicate that TRIM3 attenuates the nuclear transport of NICD by binding to Importin nuclear transport complex and retaining it in the cytoplasm; TRIM3 loss leads to enhanced NICD transport and NOTCH1 signal.

We developed a novel adult Drosophila brain tumor model by targeting brat-RNAi to neuroblasts to investigate the mechanisms related to disrupted asymmetric cellular division in stem cells (38). The penetrance of the Insc-GAL4–driven brat-RNAi was comparable with that of the Brat hypomorphic allele K06028, as determined by the size of the involved brain and the high number of actively dividing neuroblasts. One of the greatest advantages of using brat-RNAi transgene is that it can be directed to a specific cell type, such as inscuteable-positive neuroblasts, so that the other cell types are not affected. Although the adult fly brain is normally terminally differentiated and does not contain neuroblasts, the brat-RNAi model creates a phenotype dominated by proliferative neuroblastic stem cells, with a small percentage showing glial differentiation (39). We also observed the expression of stem cell markers Mira and Dpn in these proliferating neuroblasts. In our model, we were able to establish that Drosophila Brat attenuates Notch signaling and that this is accomplished at least in part by suppressing nuclear transport of NICD. As opposed to the brat mutants that develop a lethal brain tumor phenotype at the larval stage and cannot be followed longitudinally, tumor-bearing adults in the brat-RNAi model allow the investigation of the temporal evolution of mechanistic events that correspond to functional deterioration and fatal progression.

Notch is well recognized as a critical regulator of cell proliferation and self-renewal signaling pathways, and its activation is oncogenic in the proper context (1). During asymmetric division of stem cells, Notch localizes to the apical side to promote stemness, whereas Brat is a fate determinant that localizes to the basal side to drive differentiation (1). Similar to Notch pathway mutants, the loss of Brat also generates tumorous growth (8) by enabling enhanced self-renewal of neuroblasts. In this study, we observed a concrete link between Brat and Notch signaling in the adult tumor model. brat-RNAi increased the accumulation of active Notch (NICD) in neuroblasts and also led to enhanced expression of Cut, a Notch target. Genetic interaction studies in this model confirmed that suppressing the Notch pathway using Mastermind mutant MamH could reduce tumor burden in brat-RNAi flies. These observations both reestablished the role of Notch in neuroblast self-renewal and pointed to a critical contribution of Brat on the Notch signal. We also observed that brat-RNAi increases Musashi protein expression (Supplementary Fig. S6A–S6C), which, in turn, causes Numb suppression and NICD accumulation.

A recent observation in larval neuroblasts concluded that Brat and Numb regulate parallel pathways (27), rather than intersecting as we have shown in the adult fly model. The differences could be related to the effects of brat-RNAi at the adult stage as opposed to the larval stage of development. Our observation also supports the previously established Numb-independent regulation of the Notch pathway in Drosophila (28). In addition to demonstrating the regulation of Notch signaling by Brat in Drosophila, we also demonstrated that nuclear localization of NICD is suppressed by Brat, potentially providing a mechanism.

Our studies of explanted cultures of human glioblastoma indicated that the findings in Drosophila could be extended to the human disease. The human ortholog of Drosophila Brat, TRIM3, was also found to suppress NOTCH1 signaling. Unlike findings in the fly, we did not observe this strong regulation of NICD expression by TRIM3. However, analogous to Drosophila, TRIM3 was shown to attenuate the nuclear localization of NICD, and this process depended on the Importin complex. Previous studies in human cancers have suggested that ligand-independent activation of NICD may play an important role and that NICD localization inside the nucleus is a necessity for neoplastic transformation (40). After NICD is formed in early endosomes and multivesicular bodies (15, 19), nuclear localization is accomplished by its shuttling through the nuclear pore complex (NPC). Cytonuclear transportation of cargo proteins over 40 kD, such as NICD, through the NPC is generally accomplished by the Karyopherins or Importins. Proteins with a nuclear localization signal bind with Importin α, and this complex binds with Importin β to initiate the multistep process of nuclear localization (41). Previous studies in Drosophila and mammalian systems have demonstrated that NICD nuclear transport depends on the Importin complex (16–18). Here, we observed that TRIM3 bound directly to cytoplasmic Importin β1, retaining both the Importin nuclear transport complex and NICD within the cytoplasm. Our experiments also demonstrated that this retention of NICD in the cytoplasm by TRIM3 was strongly associated with attenuated stem cell properties.

On the basis of our findings, we have suggested a novel mechanism by which Brat/TRIM3 regulates Notch signaling to maintain a balance of stem cell proliferation and differentiation that is conserved across species (Fig. 5H). In Drosophila, the suppression of Brat leads to both NICD expression and its nuclear localization, whereas in human samples, it seems that the regulation of NICD by TRIM3 occurs primarily at the stage of nuclear transport. Our previous studies, as well as those of others, have indicated the negative regulation of p21 by TRIM3, a finding that could be explained by the effects of Notch signaling, as others have shown a direct regulation of p21 by Notch (5, 42, 43). Beyond this role, the regulation of nuclear transport machinery by Brat/TRIM3 could imply a far greater impact on neoplastic disease, as nuclear transport is essential for numerous functions critical to proper cell function. TRIM is a large family with many members, and here we provide evidence that at least one of its members, TRIM3, can directly bind and regulate nuclear transport in both normal and neoplastic stem cells. This regulation also seems to be evolutionarily conserved between adult Drosophila and human brain tumor models, providing rationale to pursue these mechanisms further.

No potential conflicts of interest were disclosed.

Conception and design: S. Mukherjee, D.J. Brat

Development of methodology: S. Mukherjee, R. Read, D.J. Brat

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Mukherjee, C. Tucker-Burden, C. Zhang, C. Hadjipanayis, D.J. Brat

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Mukherjee, R. Read, D.J. Brat

Writing, review, and/or revision of the manuscript: S. Mukherjee, R. Read, C. Hadjipanayis, D.J. Brat

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C. Tucker-Burden, K. Moberg, R. Read, D.J. Brat

Study supervision: S. Mukherjee, D.J. Brat

Other (technical and material support): K. Moberg

The authors thank Dr. Barry Yedvobnick (Emory University) and Dr. Cheng-Yu Lee (University of Michigan) for providing them with stocks and advice. The authors also thank Dr. Jürgen Knoblich (IMBA, Austria) for his generous gift of Brat antibody, Dr. Hideyuki Okano (Riken Brain Science Institute) for providing them with the dMusashi antibody, and Dr. Cheng-Yu Lee for his helpful comments on the manuscript.

The research reported in this article was supported in part by the Integrated Cellular Imaging Shared Resource of Winship Cancer Institute of Emory University, FACS and flow cytometry by the Emory Children's Pediatric Research Center and NIH/NCI under award number P30CA138292. This work was supported by U.S. Public Health Service NIH grants R01CA149107 and the Georgia Research Alliance (D.J. Brat).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.