Regulation of Glia Number in Drosophila by Rap/Fzr, an Activator of the Anaphase-Promoting Complex, and Loco, an RGS Protein

Abstract

Glia mediate a vast array of cellular processes and are critical for nervous system development and function. Despite their immense importance in neurobiology, glia remain understudied and the molecular mechanisms that direct their differentiation are poorly understood. Rap/Fzr is the Drosophila homolog of the mammalian Cdh1, a regulatory subunit of the anaphase-promoting complex/cyclosome (APC/C). APC/C is an E3 ubiquitin ligase complex well characterized for its role in cell cycle progression. In this study, we have uncovered a novel cellular role for Rap/Fzr. Loss of rap/fzr function leads to a marked increase in the number of glia in the nervous system of third instar larvae. Conversely, ectopic expression of UAS-rap/fzr, driven by repo-GAL4, results in the drastic reduction of glia. Data from clonal analyses using the MARCM technique show that Rap/Fzr regulates the differentiation of surface glia in the developing larval nervous system. Our genetic and biochemical data further indicate that Rap/Fzr regulates glial differentiation through its interaction with Loco, a regulator of G-protein signaling (RGS) protein and a known effector of glia specification. We propose that Rap/Fzr targets Loco for ubiquitination, thereby regulating glial differentiation in the developing nervous system.

Glia are major cell types of the nervous system and are key mediators of several neurodegenerative diseases (Kretzschmar et al. 1997; Kozuka et al. 2005; Qin et al. 2007). In the human brain, glia outnumber neurons by a factor of 10 and play a critical role in the brain's response to disease and injury (Lemke 2001; Edenfeld et al. 2005; Miller 2005). In addition to their well-known role as homeostatic regulators to provide ionic balance for neurons, glia also play a role in neuroblast proliferation, axon guidance and fasciculation, engulfment of cellular debris, and neurotransmitter uptake (Hidalgo et al. 1995; Awasaki and Ito 2004; Watts et al. 2004). Despite their widespread roles, however, the molecular mechanisms underlying gliogenesis are poorly understood. The developing nervous system of Drosophila offers a superb experimental system for examining these mechanisms. In the Drosophila embryonic and larval nervous system, glia arise from neuroblasts, the pluripotent neural stem cells that serve as precursors for all glia and neurons (Jan and Jan 2001; Chia and Yang 2002; Betschinger and Knoblich 2004; Slack et al. 2006; Yu et al. 2006; Egger et al. 2007). In the embryo, neuroblasts divide asymmetrically, giving rise to two daughter cells of unequal size. The smaller daughter cell, or ganglion mother cell (GMC), differentiates into a neuronal or a glial cell, while the larger cell is capable of self-renewal and has the potential to give rise to additional differentiated cells (Udolph et al. 1993; Campos-Ortega 1997; Edenfeld et al. 2002; Skeath and Thor 2003; Urbach et al. 2003).

Although gliogenesis and neurogenesis have been primarily studied during embryonic development, studies by Pereanu et al. (2005) show a rapid increase in glia number during the third instar larval stage. It is also during this stage that 90% of neurons in the adult CNS of Drosophila are generated (Arama et al. 2000; Bello et al. 2006, 2007). Glial differentiation during the third instar larval stage occurs in the optic stalk and glia precursor cell areas (GPC), which are located in the dorsal and ventral margins of the optic ganglion (Poeck et al. 2001; Dearborn and Kunes 2004). On the basis of position, morphology, and function, third instar larval brain glia are classified into three subtypes: (1) surface glia, (2) cortex glia, and (3) neuropile glia. Surface glia ensheath the brain and play a role in the formation of the blood–brain barrier. Cortex glia surround neurons and neuroblasts, while neuropile glia compartmentalize the neuropile, forming axon tracts for neurites (Chotard et al. 2005; Pereanu et al. 2005; Schwabe et al. 2005). In addition, another subset of glia, the retinal basal glia, are born in the optic stalk and migrate into the eye imaginal disc (Choi and Benzer 1994; Rangarajan et al. 1999). Glial cells of the optic lobe, lamina glia, epithelial glia, and the marginal glia are initially born in GPC areas and subsequently migrate into the lamina plexus (Perez and Steller 1996; Huang and Kunes 1998). Presently, specific molecular markers that distinguish these specific glia subtypes are not available; the glia-specific anti-Repo antibody is widely used to label most glial cells.

At the gene expression level, a binary mode of regulation is thought to operate within embryonic neuroblasts, wherein neuroblasts give rise to either neurons or glia. Glial differentiation requires the activation of glia-specific effector genes as well as the repression of a neuronal differentiation program (Giesen et al. 1997; Akiyama-Oda et al. 2000; Badenhorst 2001; Jones 2005). The transcription factor Glial cells missing (Gcm) has been shown to be a key regulator of glia specification. Drosophila embryos lacking Gcm function show a reduction of glia cells while ectopic expression of Gcm causes the formation of extra glia (Hosoya et al. 1995; Jones et al. 1995; Vincent et al. 1996; Bernardoni et al. 1998). Gcm activates the transcription of downstream target genes such as pointed (pnt) and reversed polarity (repo). Pnt and Repo together, in turn, promote glial cell fate specification in part by activating the transcription of the downstream effector loco (Granderath et al. 2000; Yuasa et al. 2003; Jones 2005). Thus, activation of loco is a critical step in glia differentiation. Recently, it has been shown that Gcm also functions to promote lamina neuron formation in the postembryonic larval nervous system (Chotard et al. 2005; Soustelle and Giangrande 2007). Thus, other Gcm-independent mechanisms that regulate the activation of Pnt, Repo, and Loco and the suppression of neurogenic genes such as atonal, asense, and deadpan would be critical for glial specification and terminal differentiation.

loco encodes a regulator of G-protein signaling (RGS) domain protein (Granderath et al. 1999). In mammalian systems, RGS domain proteins function as GTPase-activating proteins (GAPs) and modulate G-protein signaling by promoting GTP hydrolysis (Han et al. 2006). In Drosophila, however, Yu et al. (2005) have shown that, through its interaction with Gαi, Loco may function as a guanine nucleotide inhibitor and as a GAP. Disruption of loco gene function leads to multiple defects during development in Drosophila (Granderath et al. 1999; Pathirana et al. 2001; Yu et al. 2005). In loco loss-of-function mutant embryos, failure of glial differentiation leads to improper ensheathing of neuronal axons and, consequently, the blood–brain barrier is disrupted (Granderath et al. 1999; Schwabe et al. 2005). Biochemical studies in Drosophila embryos have shown that Loco regulates these processes by direct physical interaction with G-protein subunits (Schwabe et al. 2005). While the molecular and genetic mechanisms that control asymmetric division of Drosophila embryonic neuroblasts have been studied extensively (Jan and Jan 1998, 2001; Chia and Yang 2002; Pearson and Doe 2003; Betschinger and Knoblich 2004), the genetic components that regulate asymmetric division of neuroblasts in the postembryonic nervous system are less well understood. Cell cycle regulators and tumor suppressors, such as lethal giant larvae (Lee et al. 2006), brain tumor (Bello et al. 2006), aurora-A Kinase, polo, and imaginal disks arrested (Lee et al. 2006; Slack et al. 2006, 2007; Wang et al. 2007), have also recently been found to regulate neuroblast division during postembryonic nervous system development.

Retina aberrant in pattern/Fizzy related (Rap/Fzr) is a regulatory component of the ubiquitin ligase, anaphase-promoting complex/cyclosome (APC/C), and regulates mitotic cell cycle progression (Jacobs et al. 2002; Pimentel and Venkatesh 2005). rap/fzr mutants have a rough-eye phenotype (Karpilow et al. 1989). Here we show that in the developing larval nervous system loss of Rap/Fzr leads to an increase in glia and, moreover, targeted overexpression of Rap/Fzr leads to a dramatic decrease in glia number. Clonal analysis data derived by the mosaic analysis with repressible marker (MARCM) technique demonstrate that Rap/Fzr functions to regulate lineages of surface glia. Genetic and biochemical data presented here are consistent with a model in which Rap/Fzr regulates glia number through its interaction with Loco, an important effector of gliogenesis. We show that Rap/Fzr and Loco physically interact in tissue extracts and that Loco is ubiquitinated in vivo. We propose that Rap/Fzr targets Loco for ubiquitination, which leads to eventual proteosomal degradation. Our results reveal a novel functional role for rap/fzr and support the idea that ubiquitination, a post-translational regulatory mechanism, is essential to regulating pathways of glial differentiation.

MATERIALS AND METHODS

Fly stocks and genetic experiments:

The following mutant flies were used: w,rap3 and rap^E6 have been previously described (Karpilow et al. 1989; Jacobs et al. 2002), as has loco^P452 (Yu et al. 2005). Enhancer trap lines, P{lacZ-un1}loco^rC56w^67c23, P{lacW}rap^G0326/FM7c, and repo-GAL4/TM6;Tb, were obtained from the Bloomington Stock Center. repo-GAL4/TM6;Tb flies were used to drive expression of UAS-rap/fzr, UAS-loco-GFP (gift from Ulrike Gaul). rap/fzr mosaic clones were generated using FLP-mediated mitotic recombination (Xu and Rubin 1993). rap^G0326P{neoFRT}18A/FM7 or rap/fzr^8F3 P{neoFRT}19A/FM7 recombinants (a gift from Christian Klambt) were crossed to P{neoFRT}18A arm-lacZ; hsFLP (a gift from Ting Xie) and P{neoFRT}19A arm-lacZ; hsFLP (a gift from Nick Baker).

MARCM clones were generated as described (Lee and Luo 1999), and the following fly stocks were used: P{neoFRT}19A, P{tubP-GAL80}LL1, P{hsFLP}1, w[*]; P{UAS-mCD8∷GFP.L}LL5/+;repoGAL/+, P{neoFRT}19A arm-lacZ/+, and UAS- rap/fzr/+. For MARCM and FLP/FRT-mediated recombination, first and second instar larvae were collected in separate vials and heat pulsed at 37° two times at 2-hr intervals (Slack et al. 2006). Eye disc–brain complexes from female third instar larvae were dissected for all mosaic experiments.

Immunohistochemistry:

Whole-mount preparations of third instar larval brains were stained with the appropriate antibodies and counterstained with fluorescent secondary antibodies as described previously (Pimentel and Venkatesh 2005). Anti-Repo mouse (8D12) and anti-Dachsund mouse (Mabdac1-1) supernatants were obtained from Developmental Studies Hybridoma Bank (University of Iowa) and used at a dilution of 1:5 for all experiments. Anti-phospho-histone H3 rabbit antibody from Upstate Biotechnology (Lake Placid, NY) was diluted 1:100 for experiments. Anti-Loco C1 antibody (a gift from William Chia) was used at a dilution of 1:250. Anti-Miranda antibody (a gift from Fumio Matsuzaki) was used at a 1:1000 dilution. Anti-Fzr (human) antibody from Invitrogen (San Diego) was used at a dilution of 1:10. Anti-β-galactosidase (rabbit) antibody from eBioscience was used at a dilution of 1:50. Polyclonal rabbit anti-GFP antibody (Abcam) was used at 1:250 dilution.

Protein isolation and Western blot analysis:

To prepare tissue extracts for protein lysates, UAS-loco-GFP;repo-GAL4 larvae were homogenized in immunoprecipitation buffer (10 mm Tris, pH 7.5, 80 mm glycerophosphate, pH 7.3, 20 mm EGTA, pH 8.0, 15 mm MgCl₂, 0.5 mm DTT, 2 mm Na₃VO₄, 10% glycerol, 0.1% NP40). Protein lysate (500 μg) was precleared using 50 μl of protein IgG beads (eBioscience). Precleared lysate (500 μl) was incubated overnight at 4° with 1 μl of Loco C1 antibody, 1 μl of anti-GFP (Abcam), or 1 μl of anti-Rap/Fzr (Pimentel and Venkatesh 2005) and 50 μl of protein IgG beads. Beads were washed three times with immunoprecipitation buffer before addition of 50 μl of SDS sample buffer (250 mm Tris, 8.2% SDS, 05% bromophenol blue, 40% glycerol, and 200 mm of DTT). Immunoprecipitated protein was extracted by boiling protein IgG beads in SDS sample buffer for 5 min.

Proteins from larval extracts were separated by SDS–PAGE using a 12% resolving gel and a 5% stacking gel. Proteins from the SDS–PAGE gel were transferred to a 0.45-μm nitrocellulose membrane (Pierce, Rockford, IL) for 1 hr at 100 V at 4°. Membranes were blocked in 5% skim milk dissolved in Tris-buffered saline containing 0.2% Tween 20 (TBS-T) for 2 hr and were incubated overnight in primary antibodies, using a 1:1000 dilution for polyclonal anti-GFP antibody, polyclonal anti-ubiquitin antibody, and anti-Loco C1 antibody. Polyclonal anti-Rap/Fzr antibody was used at a 1:150 dilution. After primary antibody incubations, blots were washed three times in TBS-T for a period of 5 min for each wash. Membranes were then incubated for 1 hr in secondary antibody using a 1:10000 dilution of anti-rabbit IgG–horseradish peroxidase (Jackson Labs) and a 1:1000 anti-rabbit Trueblot antibody (eBioscience), each diluted in blocking solution. After secondary antibody incubation, membranes were washed for 10 min three times in TBS-T. For detection of horseradish peroxidase on immunoblots, an ECL Western blotting substrate kit (Pierce) was used according to the manufacturer's protocol. The very low endogenous levels of Loco protein in wild-type larval tissue resulted in extremely weak signals in our immune precipitation and Western blot experiments. To obtain stronger signals, we elevated the levels of Loco expression by using tissue extracts from transgenic Drosophila strains expressing UAS-loco-GFP. Expression of Loco-GFP in glia was achieved using the repo-GAL4 as a driver, and anti-GFP antibody was used to detect Loco in all larval immunoprecipitation and Western blot experiments.

Statistics and quantitative analysis of glia and neurons:

Third instar larval brains stained with specific antibodies were examined in a LSM 510 (Zeiss) confocal microscope. The following antibodies were used for labeling various cell types: glia cells (anti-Repo), neuroblasts (anti-Miranda), mitotic cells (anti-phosphohistone-H3), neurons (anti-Dachshund), and apoptotic cells (anti-Caspase). To account for the three-dimensional nature of the larval brain, images were obtained as confocal Z stacks and quantification of the cell numbers was done from maximum projections. For quantification of cell numbers, we employed two different types of image analysis software. The Volocity image analysis software (Improvision, Waltham, MA) was used for counting cells in three-dimensional projections. Image J software (http://rsb.info.nih.gov/ij/) was used for counting glia and neurons located on the surface of the brain. The top optical sections (5 μm below the surface of the brain) from confocal Z-stacks were used for analysis of surface glia. Significance and P-values were derived using Student's t-test. For high-intensity fluorescent images involving densely packed tissue, the Volocity software uses voxel intensity measurements to quantify cell number (Cuschieri et al. 2006). For larval brains stained with anti-Dachshund antibodies, confocal image stacks were acquired using the same settings for all samples, with a gain of 815.

RESULTS

Rap/Fzr regulates glia differentiation:

To test whether Rap/Fzr plays a role in glia differentiation during the third instar larval stage, we stained third instar larval brains from rap/fzr loss-of-function mutants (w,rap3) with the glia-specific marker anti-Repo antibody (Figure 1B). We found a significant increase in glia number in the rap/fzr loss-of-function mutant (w,rap3) compared to wild type (Canton-S) brains (Figure 1D; P < 0.028, n = 17). Conversely, to study the effects of ectopic expression of Rap/Fzr, we employed the UAS-GAL4 system (Brand and Perrimon 1993) and expressed UAS-rap/fzr in glia using the repo-GAL4 driver. UAS-rap/fzr; repo-GAL4 larval brains showed a significant decrease in glia number (Figure 1, C and D; P < 0.00000014; n = 17) compared to wild type (Figure 1A). Since rap/fzr expression is primarily restricted to surface glia (see below), we examined changes in surface glia number in rap/fzr mutants. Since markers that specifically label different glia subtypes are not currently available, we identified surface glia on the basis of their position in the larval brain (see materials and methods). Glia located on the superficial regions of larval brains were analyzed in rap/fzr mutants. rap/fzr loss-of-function mutants (w,rap3) showed a significant increase in surface glia number compared to wild-type flies (Figure 1E; P < 0.002, n = 13). Conversely, overexpression of rap/fzr in larval brains (UAS-rap/fzr; repo-GAL4) resulted in a significant decrease in glia number (Figure 1E; P < 0.0002, n = 14) compared to wild type.

Figure 1.—

Rap/Fzr regulates glial differentiation. (A–C) Confocal images of third instar larval brains stained with the glia-specific marker Repo (green) and the mitotic marker anti-phospho-histone H3 (red). (B) rap/fzr loss-of-function mutants, w,rap3, show a significant increase in glia number compared to (A) wild type (glia number for w,rap3 is 3431 cells ± 375; for wild type, 2390 cells ± 121, n = 17; P < 0.028). (C) Ectopic expression of rap/fzr shows a decrease in glia number (glia number for UAS-rap/fzr; repo-GAL4 is 724 cells ± 128, n = 17; **P < 0.00000014). (D) Histograms of glia number within the entire brain of third instar larvae. (E) Quantification of surface glia cells in wild type, w,rap3, and UAS-rap/fzr; repo-GAL4 larval brains (wild type, 380 ± 25; w,rap3, 682 ± 59; UAS-rap/fzr; repo-GAL4, 148 cells ± 35; n = 22, n = 21, P > 0.25, and P > 0.12; n = 13, n = 14, *P < 0.002, and **P < 0.0002). (F) Histogram of mitotic index quantified on the basis of phospho-H3 labeling of larval brains from w,rap3 and UAS-rap/fzr; repo-GAL4. No significant difference in the mitotic index was observed between wild type (1086 ± 117); w,rap3 (1621 ± 322); UAS-rap/fzr; repo-GAL4 (1554 cells ± 256; n = 22, n = 21, P > 0.25 and P > 0.12). Images are maximum projections taken from a LSM 510 confocal microscope. Bar, 50 μm.

To test whether the observed increase in glia number was due to an increase in mitosis, we stained larval brains with the mitotic marker anti-phospho-histone H3 antibody and estimated the mitotic index. We did not detect significant changes in the mitotic index of third instar larval brains in w,rap3 mutants (Figure 1F; P > 0.25, n = 22). Similarly, the brains of the gain-of-function (UAS-rap/fzr; repo-GAL4) animals do not show a change in the mitotic index compared to wild-type brains (Figure 1F; P > 0.12, n = 21).

Rap/Fzr promotes neuron formation:

The binary model of neuron glia differentiation predicts that a given GMC differentiates into either a neuron or a glial cell (Hosoya et al. 1995; Jones et al. 1995; Vincent et al. 1996). To test whether the changes that we observed in the glia number in rap/fzr mutants were accompanied by corresponding changes in the number of neurons, we first quantified the total number of neurons in larval brains. Loss-of-function rap/fzr mutants (w,rap3) showed a significant decrease in the number of neurons as evidenced by anti-Dachshund (neuron marker) staining compared to wild type (Figure 2, A, B, and D; P < 0.017, n = 17). Conversely, ectopic overexpression of rap/fzr in third instar larval brains (UAS-rap/fzr; repo-GAL4) showed a significant increase in the number of neurons compared to wild-type larval brains (Figure 2, A, C, and D; P < 0.0003, n = 15). In addition, we determined the number of neurons in the region of the brain corresponding to the surface glia location in rap/fzr mutant larvae. Loss-of-function mutants showed a significant decrease in surface neurons compared to wild-type flies (Figure 2E; P < 0.05, n =12). Consistent with our data on surface glia, UAS-rap/fzr; repo-GAL4 larval brains showed a significant increase in surface neurons compared to wild-type larval brains (Figure 2E; P < 0.05, n = 16). These observed changes in the neuronal number are consistent with the binary switch model for neuron and glia differentiation (Hosoya et al. 1995; Jones et al. 1995). Consistent with this model, the expression of Rap/Fzr effects a switch in which a precursor cell differentiates into a neuron instead of glial cell.

Figure 2.—

Rap/Fzr promotes neuron formation. Third instar larval brains were stained with the early neuronal marker Dachshund (green) and the mitotic marker anti-phospho-histone H3 (red). Compared to wild-type brains (A), rap/fzr loss-of-function mutants w,rap3 (B) show a significant decrease in neuron density. (C) Ectopic overexpression of Rap/Fzr in brains results in an increase in neuron density. (A–C) Single optical sections from a confocal microscope. (A′–C′) Magnified areas of the brain (outlined squares in A–C) show densely packed neurons within third instar larval brains. (A″–C″) Quantification of surface neurons within the brain by Image J software. Projections with numerous colors represent individual cells that were counted. (D) Histogram of neuron density measured by the Volocity software (w,rap3, 1.08 × 10⁶ voxels ± 1.40 × 10⁵; wild type, 1.62 × 10⁶ voxels ± 1.41 × 10⁵, n = 17, *P < 0.017; UAS-rap/fzr; repo-GAL4, 3.07 × 10⁶ voxels ± 2.38 × 10⁵, n = 15, **P < 0.0003). (E) Histogram shows quantification of surface neurons (wild type, 398 neurons ± 33.3; w,rap3, 300 neurons ± 30.1; UAS-rap/fzr;repo-GAL4, 488 neurons ± 38.44, n = 12, n = 16, *P < 0.05 and *P < 0.05). Bar, 50 μm.

Rap/Fzr regulates in glia differentiation in a cell-autonomous manner:

The rap/fzr loss-of-function mutants (w,rap3) used above are strong hypomorphs that nonetheless express residual levels of Rap/Fzr activity (A. H. Korayem, R. Chako and T. Venkatesh, unpublished results). To ascertain the phenotypic effects of a complete loss-of-function mutant and to study whether rap/fzr controls glia development in a cell-autonomous manner, we generated mosaic tissue with clones of rap/fzr− tissue surrounded by wild-type tissue, using the FLP/FRT technique (Xu and Rubin 1993; Golic et al. 1997). The rap/fzr null allele (rap/fzr^8F3) was recombined into a FRT chromosome carrying the armadillo-lacZ (arm-lacZ) as a reporter. Following mitotic recombination, rap/fzr− mutant clones were identified by the lack of β-galactosidase expression, while the surrounding wild-type clones showed expression of arm-β-galactosidase. Following staining with anti-β-galactosidase and anti-Repo antibodies, larval brains were analyzed by confocal microscopy. The rap/fzr− mutant patches that lacked arm-lacZ expression showed a significant increase in the number of glia (Repo-positive cells) (Figure 3, A, B, and D; P < 0.004, n = 10) compared to the adjacent wild-type tissue (Figure 3C), suggesting that rap/fzr regulates glia number in a cell-autonomous manner.

Figure 3.—

Rap/Fzr functions cell autonomously to regulate glia number. Confocal microscope images of third instar larval brains stained with anti-β-galactosidase (red) and anti-Repo (green). (A) An example of a rap/fzr− clone located within the larval brain marked by the lack of anti-β-galactosidase (red) staining and a white outline. The corresponding wild-type tissue is outlined in yellow. (B) Magnified image of the outlined area from the rap/fzr− clone in A. (C) Magnified image of wild-type tissue shown as yellow inset in A. (D) Histogram shows that the number of glial cells in rap/fzr-- clones are increased compared to wild-type tissue (rap/fzr− clones, 19 glia cells ± 2.7; wild type, 8 glia cells ± 1; n = 10; *P < 0.004). rap/fzr mosaic clones were induced by the FLP/FRT technique (see materials and methods for details). Bar, 10 μm.

Apoptosis does not contribute to change in glia number:

To test whether the significant decrease in glia number in UAS-rap/fzr;repo-GAL4 animals was due to apoptosis, we stained third instar larval brains with the apoptosis marker anti-Caspase-3 (Baker and Yu 2001). Our data showed that the reduced glia number in larval brains following ectopic overexpression of rap/fzr did not correlate with increased apoptosis (see supplemental Figure S1, A–D; P > 0.85, n = 10). Similarly, in loss-of-function rap/fzr mutants, where glia numbers increase there was not a significant change in the levels of apoptosis as defined by anti-Caspase-3 staining (see supplemental Figure S1, A–D; n = P > 0.33, n = 11). Taken together, these data suggest that apoptosis does not significantly contribute to the change in glia number observed in rap/fzr mutants.

Because ectopic overexpression of rap/fzr in glia caused a dramatic reduction in glial cell number, we sought to test whether this result was cell type specific or whether the expression of rap/fzr in non-glia cell types could lead to similar results. We expressed UAS-rap/fzr in other cell types employing a variety of other GAL4 drivers. Ectopic expression of rap/fzr using other GAL4 drivers had no effect (data not shown). These results suggest that the changes in glia number observed with changes in the expression of Rap/Fzr in the larval brain are not merely nonspecific effects.

MARCM analysis reveals that rap/fzr specifically regulates surface glia lineages:

Our results presented above show that, in larval brains, loss-of-function of rap/fzr generates a significant increase in glia cell number during third instar larval stage. Conversely, ectopic overexpression of rap/fzr results in the reduction of glia number. To test whether rap/fzr regulates glia fate at the single-cell level, we used MARCM (Lee and Luo 1999). MARCM technique incorporates both FLP/FRT and UAS/GAL4 systems, facilitates the tracking of cell lineages, and is particularly useful for generating gain-of-function clones. In our experiments, we tracked cells overexpressing rap/fzr as MARCM clones marked by the expression of mcD8-GFP. To ascertain the effects of rap/fzr on glial cell fate, we examined whether the mcD8-GFP-positive MARCM clones were also Repo positive. Interestingly, when rap/fzr is overexpressed within surface glia, only 26% of MARCM clones were Repo positive (Figure 4, A–C). However, when we examined the optic lobe or deeper regions of the larval brain such as neuropile lineages, 81% of the MARCM clones were still Repo positive (Figure 4, D–F), suggesting that rap/fzr regulates surface glia and has no effect on the glia from the interior regions (neuropile and optic lobe). We analyzed 31 clones from eight larval brains and our results are consistent with the model that overexpression of rap/fzr suppresses the formation of surface glia. These data together with the mosaic analysis of rap/fzr− null clones show that loss of rap/fzr function promotes glia cell number while gain of function of rap/fzr suppresses the generation of surface glia.

Figure 4.—

Rap/Fzr regulates development of surface glia. Single optical sections from a confocal microscope show MARCM clones that are GFP positive (green). (A–C) Examples of rap/fzr overexpression clones (arrows) showing that the majority of the clones located on the surface do not contain glial cells (red, anti-Repo); only 26% are Repo positive. (D–F) rap/fzr overexpression clones located within the neuropile region of the brain show that 81% are Repo positive. Bar, 50 μm.

Rap/Fzr genetically interacts with Loco, a positive effector of glia differentiation:

Our genetic modifier studies identified Loco as a dominant suppressor of the rap/fzr rough-eye phenotype (Kaplow et al. 2007). Loco plays a critical role in embryonic asymmetric cell division of neuroblasts during gliogenesis and also mediates septate junction formation during glia differentiation (Schwabe et al. 2005; Yu et al. 2005). To test whether Loco interacts with Rap/Fzr to regulate glia differentiation in the larval nervous system, we analyzed third instar larval brains of the loss-of-function allele (loco^p452). A significant decrease in glia number was seen in the loss-of-function mutant loco^p452/loco⁴⁵², compared to wild type (Figure 5, A, B, and E; P < 0.002, n = 15). This loco^p452/loco⁴⁵² phenotype is similar to the glia phenotype of the rap/fzr gain of function (UAS-rap/fzr; repo-GAL4). Our genetic interaction studies showed that a single copy of the loco^p452 mutation acts as a dominant suppressor of the rough-eye phenotype. As with w,rap3 mutants, the larval brains of a second rap/fzr allele, rap^E6/fzr (a weak hypomorphic allele) animals, show a significant increase in glia number compared to wild-type brains (Figure 5, B, C, and E; P < 0.02, n = 15). To examine whether loco^p452 could suppress the rap^E6/fzr phenotype, we generated rap^E6/fzr; loco^p452/+ animals. A single copy of the loco^p452 mutation was able to suppress the rap^E6/fzr glia phenotype. Larval brains from rap^E6/fzr; loco^p452 /+ show glia numbers similar to wild type (Figure 5, B, D, and E). The larval brains of the heterozygote loco⁴⁵²/+ have wild-type glia number (data not shown). These results are consistent with a model in which rap/fzr and loco genetically interact in a dosage-sensitive manner to regulate glia number in the larval brain.

Figure 5.—

rap/fzr and loco genetically interact to regulate glial differentiation. Confocal images of third instar larval brains stained with the glia-specific marker Repo. Compared to wild-type brains (B), loco^P452, a loss-of-function mutant (A), shows a significant decrease in glia number (1628 ± 161 Repo-positive cells compared to 2390 ± 126 for wild type, n = 15, *P < 0.002). Conversely, the hypormorphic allele rap^E6 (C) shows a marked increase in glia number compared to wild-type brains (3627 ± 388 Repo-positive cells, *P < 0.02, n = 15). The rap^E6 glia phenotype is suppressed with (D) one copy of loco^P452 (brains analyzed for rap^E6; loco^P452/+ show 2464 ± 175 Repo-positive cells; n = 12, *P < 0.03). Images are maximum projections from a confocal microscope. Bar, 50 μm. (E) Histogram of glia number in third instar larval brains.

Rap/Fzr and Loco physically interact:

During cell cycle, Rap/Fzr directly interacts with cellular substrates to target them for ubiquitination by the ubiquitin ligase APC/C. The interaction between Rap/Fzr and cellular substrates requires either a KEN box or a Destruction box motif (D box: RXXLXXXN) (Pfleger and Kirschner 2000). Bioinformatic analyses of Loco amino acid sequence showed both a KEN box and a Destruction box, raising the possibility that Loco is a potential substrate of Rap/Fzr (Kaplow et al. 2007). To test whether Rap/Fzr and Loco physically interact in vivo, we performed a series of co-immunoprecipitation experiments with larval brain lysates from UAS-loco-GFP; repo-GAL4. First, we immunoprecipitated Loco-GFP and probed with anti-Rap/Fzr antibody (Figure 6A). We further performed a reciprocal experiment in which Rap/Fzr was immunoprecipitated with anti-Rap/Fzr and probed with anti-GFP antibody (Figure 6B). Our results showed that Rap/Fzr and Loco are found in a complex and suggest that Rap/Fzr and Loco physically interact in vivo. Thus, Loco is a likely substrate for ubiquitination by the APC/C.

Figure 6.—

Biochemical evidence shows that Loco interacts with Rap/Fzr and that Loco is ubiquitinated in vivo. (A) Protein lysates from the UAS-loco-GFP; repo-GAL4 third instar larval brain–eye complex were used to immunoprecipitate Loco-GFP protein. Western blot analysis using anti-GFP reveals a 100-kDa band corresponding to Loco protein in larvae. Negative control shows the absence of Loco protein (left). Rap/Fzr co-immunoprecipitates with Loco in the third instar larval brain–eye complex. A 53-kDa band is detected using Rap/Fzr antibody in larval lysates. The negative control shows the absence of Rap/Fzr when IgG beads were incubated without GFP antibody (right). (B) Protein extracts from UAS-loco-GFP; repo-GAL4 third instar larval brain–eye complex were used to immunoprecipitate Rap/Fzr protein. Western blot analysis using anti-Rap/Fzr reveals a 53-kDa band corresponding to Rap/Fzr protein in larval brain extracts (left). The negative control shows the absence of Rap/Fzr protein. Data in the right panel show that Loco physically interacts with Rap/Fzr in the third instar larval brain–eye complex. A 100-kDa band is detected using anti-GFP antibody in larvae, representative of Loco protein. The negative control shows the absence of Loco protein when IgG beads were incubated without Rap/Fzr antibody. (C) Loco is ubiquitinated in vivo. Protein lysates from UAS-loco-GFP; repo-GAL4 third instar larval brains were again used to immunoprecipiate Loco-GFP protein. Western blot analysis using anti-GFP reveals a 100-kDa band corresponding to Loco protein in larval extracts (left). The negative control shows the absence of Loco protein. Western blot analysis with GFP-Loco immunoprecipitate reveals the presence of ubiquitin in 80- and 100-kDa bands (right). The inset (bottom right) represents lighter exposure to visualize bands clearly.

To test whether Loco is ubiquitinated in vivo, third instar larval brain lysates from UAS-loco-GFP; repo-GAL4 larvae were immunoprecipitated with anti-GFP antibodies and probed with anti-ubiquitin on Western blots (Figure 6C). Two protein bands of 100 and 80 kDa were detected and these correspond to the Loco protein isoforms, suggesting that they are ubiquitinated (Figure 6C). Taken together, these results are consistent with a model in which Rap/Fzr and Loco physically interact and target Loco for ubiquitination in vivo. Furthermore, our model predicts that changes in rap/fzr gene dosage should be reflected in the cellular levels of Loco protein. We thus examined Loco levels in larval brains of wild-type and rap/fzr loss-of-function and gain-of-function strains using antibody staining techniques. Immunofluorescence data show higher levels of Loco in rap/fzr loss-of-function mutants compared to wild type (see supplemental Figure S2). Conversely, UAS-rap/fzr; repo-GAL4 larval brains show weak Loco expression compared to wild-type larval brains. These results further support the model that Rap/Fzr directly targets Loco for ubiquitination followed by proteosomal degradation.

Rap/Fzr and Loco are expressed in a subset of glia in the developing larval brain and eye imaginal disc:

Our model also predicts that Rap/Fzr and Loco are expressed in the same subset of glial cells. To test whether Rap/Fzr and Loco are coexpressed in glia cells, we monitored expression patterns of Rap/Fzr and Loco in the third instar larval brains and eye imaginal discs. We used two enhancer trap lines with P-element insertions, rap/fzr^P2241 and loco^rC56, respectively, and examined the tissue expression patterns of both rap/fzr and loco. Third instar larval brains from rap/fzr^P2241 and loco^rC56 were double-labeled with anti-β-galactosidase and the glia-specific anti-Repo antibody. Results from rap/fzr^P2241 staining showed strong rap/fzr expression on the periphery of the brain, localized to surface glia (see supplemental Figure S3). Robust rap/fzr expression was also observed in cells in the deeper regions of the brain but the lack of Repo staining in these cells suggests that these are non-glial cells (see supplemental Figure S3). Unlike rap/fzr^P2241, which showed expression predominantly localized to surface glia, loco enhancer trap line loco^rC56 showed expression within most glia cells, including surface, cortex, and neuropile glia (see supplemental Figure S4).

On the basis of our genetic and biochemical data, we expected colocalization between Rap/Fzr and Loco within third instar larval brains. Double labeling with antibodies against Drosophila Loco and human anti-Rap/Fzr/Cdh1, followed by confocal microscopy, showed colocalization of Loco (green) and Rap/Fzr (red) on the surface of the brain (see supplemental Figure S5). Strong colocalization between Rap/Fzr and Loco is especially prominent along the ventral edge of the brain near the margin between the brain and ventral ganglion. However, Loco and Rap/Fzr distribution within deeper layers of the larvae brain is disparate. Within the cortex, Rap/Fzr expression is present in the optic lobe and in a dispersed cluster of cells on the ventral side of the brain. Loco staining within the cortex is relatively weak compared to Rap/Fzr staining. Only a few cells on the ventral side of the brain are Loco positive, and in the optic lobe, Loco expression is virtually undetectable (see supplemental Figure S5).

To test whether both proteins are expressed within glia cells, wild-type larval brains were triple-labeled with anti-Loco (blue), anti-Rap/Fzr (red), and anti-Repo (green) antibodies. Our results showed that, in the surface regions of the brain, Rap/Fzr and Loco are localized to the cytoplasmic region of the glia whose nuclei are stained with anti-Repo (Figure 7, A–E, arrow). When localization of Rap/Fzr, Loco, and Repo was analyzed in the deeper layers of the brain, distribution of Loco was very diffuse within the cortex. Strong localization of Rap/Fzr was seen within the optic lobe surrounding marginal, lamina, and epithelia glia (Figure 7, A′–E′).

Figure 7.—

Loco and Rap/Fzr colocalize within a subset of glial cells in third instar larval brains. Single optical sections from a confocal image show Loco (blue), Rap/Fzr (red), and Repo (green) distribution within the exterior surface of the larval brain (A–C). A merged image shows triple-labeled brain with Rap/Fzr (red), Loco (blue), and Repo (green) localized within a subset of glia cells (D). Loco and Rap are localized within the cytoplasm of glia indicated by the arrow in E. (A′–C′) Deeper regions of the brain. Loco (blue) expression is diffuse in the brain's interior and is restricted to the ventral region of the brain. Unlike Loco, Rap/Fzr (red) expression is apparent in two regions, the optic lobe and the ventral edge of the brain. Images in D and D′ are a merge of images in A–C and A′–C′, respectively. Squares in D and D′ outline magnified regions displayed in E and E′, respectively. Bar, 50 μm.

Rap/Fzr and Loco protein are localized within third instar larval neuroblasts:

In addition to their expression in surface glia, Rap/Fzr and Loco are also expressed in a subset of cells that are not Repo positive. Because Loco has been previously shown to play a role in asymmetric division of neuroblasts (Yu et al. 2005), we tested whether these Repo-negative cells represented either neuroblasts or GMCs. We used the GMC marker anti-Prospero and the neuroblast marker anti-Miranda (Ceron et al. 2006) to label neuroblasts and GMCs located in third instar larval brains. Although Prospero shows no colocalization with Rap/Fzr or Loco, both Rap/Fzr and Loco are expressed in a subset of Miranda-positive neuroblasts (Figure 8, A–D), suggesting that Rap/Fzr and Loco also function early in neuroblasts during the third instar larval stage. In addition, rap/fzr loss-of-function (w,rap3) mutants showed no significant differences in neuroblast number compared to wild type as monitored by Miranda staining (Figure 9, A and B; P > 0.100, n = 15). By contrast, ectopic overexpression of Rap/Fzr in glial cells resulted in a significant increase in the number of neuroblasts as evidenced by the increased number of Miranda-positive cells (Figure 9C; P < 0.002, n = 15). These data are consistent with a role for Rap/Fzr and Loco during early stages of neuroblast determination as well as later during glia differentiation.

Figure 8.—

Rap/Fzr and Loco are localized within neuroblasts located on the brain surface. Third instar larval brains showing immunolocalization of Loco (blue), Rap/Fzr (red), and Miranda (green) within neuroblasts in the superficial areas (A–C). A merge of the images A–C shows colocalization of Loco and Rap/Fzr in Miranda-positive cells (D); the square in D indicates the magnified area displayed in E. Bar, 50 μm. Images shown are single optical sections from a confocal microscope.

Figure 9.—

Overexpression of rap/fzr generates extra neuroblasts in third instar larvae brains. Confocal images of third instar larval brains were labeled with the neuroblast marker Miranda (red) and the glial marker Repo (green) antibody. Compared to wild type (A), rap/fzr gain-of-function UAS-rap/fzr; repo-GAL4 (C) brains show a significant increase in Miranda-positive cells (5416 neuroblasts compared to 3067 neuroblasts in normal larval brains, n = 15, *P < 0.002). Student t-test reveals no significant difference in neuroblast numbers between the loss-of-function mutant w,rap3 (B) and wild type (4356 neuroblast in w,rap3 and 3067 neuroblasts normal larvae, n = 15; P < 0.100). Arrows show some Repo-positive cells that are also Miranda positive in wild type, w,rap3, and UAS-rap/fzr;repo-GAL4 larvae. (D) Bar graph represents quantification of neuroblasts using Volocity software. Images are maximum projections taken from a Zeiss LSM 510 confocal microscope. Bar, 50 μm.

To examine whether the ectopic neuroblasts formed in UAS-rap/fzr; repo-GAL4 larvae had the potential to self-renew and give rise to neurons, wild-type, UAS-rap/fzr; repo-GAL4, and w,rap3 larvae were pulse fed with bromodeoxy uridine (BrdU) for 15 hr at 24, 48, and 96 hr post-larval hatching. UAS-rap/fzr; repo-GAL4 larvae incorporated a significant amount of BrdU, while in rap/fzr loss-of-function mutants, BrdU incorporation was similar to wild-type larvae (supplemental Figure S7). These data suggest that the ectopic neuroblasts formed in rap/fzr gain-of-function (UAS-rap/fzr; repo-GAL4) larval brains are proliferative and maintain the potential to give rise to neurons.

DISCUSSION

The APC/C is a multi-subunit ubiquitination complex that has been well characterized for its role in regulating mitotic exit (Sudakin et al. 1995; Geley et al. 2001). Rap/Fzr/Cdh1 is an activator of APC/C and plays a key role in the regulation of mitosis by targeting cell cycle regulators, such as cyclins and cyclin-dependent kinases, for ubiquitination (Schwab et al. 1997; Sigrist and Lehner 1997; Pimentel and Venkatesh 2005). Our data uncover a novel role for Rap/Fzr in the regulation of glia differentiation. Loss-of-function rap/fzr mutants display an increase in glia number and a corresponding decrease in neuronal number. Conversely, targeted overexpression of Rap/Fzr in glia leads to a severe reduction in glia number with a corresponding increase in neuronal number. This change in glia and neuron number occurs without significantly altering the mitotic index. Similarly, Pereanu et al. (2005) reported a change in glial cell number in the larval brain without a significant change in mitotic index and suggested that the additional glial cells arise from the differentiation of secondary neuro-glioblasts located in the surface of the brain. Our clonal analysis data derived from MARCM experiments suggest that Rap/Fzr specifically regulates differentiation of a subset of glia, the surface glia. Several lines of evidence presented here support the model that Rap/Fzr regulates gliogenesis by targeting the RGS protein, Loco, for ubiquitination. First, our genetic interaction studies show that a single copy of the loco mutation is a dominant suppressor of both the rap/fzr rough-eye phenotype and the glial phenotype in the larval brain. Second, our biochemical data show an interaction between Rap/Fzr and Loco in larval brain tissue and that Loco is ubiquitinated in larval extracts. Third, results from immunolocalization experiments show that Rap/Fzr and Loco colocalize within surface glia in the postembryonic larval brain. We conclude that Loco is targeted for ubiquitination by Rap/Fzr through its D-box and/or KEN box, two signature ubiquitination-targeting motifs recognized by the APC/C (Pfleger and Kirschner 2000; Burton and Solomon 2001; Hilioti et al. 2001).

Loco has been previously shown to be a positive effector of glia development during Drosophila embryogenesis (Granderath et al. 1999, 2000; Granderath and Klambt 2004). Recently, Loco has also been reported to have a role during the asymmetric cell division of embryonic neuroblasts (Yu et al. 2005). Our results suggest a new role for Loco in postembryonic development of Drosophila CNS and, specifically, in glial differentiation. We propose (Figure 10) that the cellular level of Loco in the postembryonic GMC is a key positive effector in the binary switch model of glia–neuron differentiation. In this model, Rap/Fzr negatively regulates glia number by targeting Loco for ubiquitination and eventual proteosomal degradation. Our model further predicts that alteration in the rap/fzr gene dosage would change cellular levels of Loco, with resulting effects on glia number. In larval neuroblasts, compartmentalization of Loco within GMCs may be critical in promoting a glial cell lineage. Our results showed that, in the larval brain, Loco is colocalized with Miranda and Rap/Fzr in the basal axis, whereas during asymmetric division of embryonic neuroblasts, Loco is expressed in the apical axis (Yu et al. 2005). Although Miranda is a known mediator of asymmetric division of embryonic neuroblasts and a specific marker for larval neuroblasts, its function in postembryonic development has not been completely elucidated (Ikeshima-Kataoka et al. 1997; Shen et al. 1998; Slack et al. 2006). Colocalization of Loco with Miranda and Rap/Fzr (Figure 7) suggests a possible functional role for these molecules during postembryonic neuroblast division.

Figure 10.—

A proposed model for the interaction between Rap/Fzr and Loco. (A) In wild-type third instar larvae, neuroblasts located on the exterior surface of the brain divide asymmetrically, creating a larger daughter cell, a neuroblast (NB, yellow), and a smaller daughter cell, the GMC (blue). The GMCs will divide once to give rise to either two neurons (red) or two glia (green). Rap/Fzr and Loco are localized within the neuroblasts and function to control neuroblast and glia number. Rap/Fzr targets Loco for ubiquitination by APC/C and regulates its levels. (B) rap/fzr loss-of-function leads to elevated levels of Loco. Consequently, Loco accumulation within the GMCs leads to an increase in glia number. When Rap/Fzr is overexpressed (C) during the third instar larval stage, Loco is targeted for degradation, leading to an increase in neuroblast number. Furthermore, destruction of Loco within the GMCs leads to the formation of neurons.

Collectively, our data suggest that Rap/Fzr regulates glia differentiation during two phases of development: (1) initially, Rap/Fzr controls the proliferation and self-renewal of dividing neuroblasts, and (2) subsequently, Rap/Fzr regulates the differentiation of GMCs. This model is consistent with evidence from other studies showing that proliferation of larval neuroblasts is controlled by other components of the APC/C (Slack et al. 2006), such as ida (a subunit of the APC/C), and Aurora-A kinase (Lee et al. 2006; Wang et al. 2006; Slack et al. 2007), a known target of APC/C-mediated ubiquitination during mitotic progression (Littlepage and Ruderman 2002). Since work by Slack et al. (2006) has shown a possible role for ida and, in turn, for the APC/C during neuroblast division, it would be interesting to determine if additional components of the APC/C have roles during later phases of development. Our preliminary analysis of glia number in morula/APC2 (a catalytic subunit of APC/C) mutants showed a significant increase in glia number (data not shown) similar to rap/fzr loss-of-function mutants. However, the precise roles of additional components of the APC/C, a complex of 11 subunits, during glial differentiation have yet to be elucidated. While our results suggest that Rap/Fzr regulates neuroblast number by targeting Loco for degradation, Rap/Fzr may also regulate neuroblast self-renewal through its interactions with other proteins such as Aurora-A kinase. In Drosophila larval neuroblasts, Aurora-A kinase is an important regulator of neuroblast self-renewal (Lee et al. 2006; Wang et al. 2006) and is known to be a substrate for APC/C in vertebrates (Littlepage and Ruderman 2002; Stewart and Fang 2005, Feine et al. 2007) .

The data presented in this article support a model in which components of the ubiquitin ligase complex, APC/C, mediate a post-translational regulatory mechanism critical to the glial differentiation program. During the past 2 years, other studies have also reported novel roles for the APC/C and its components during nervous system development, independent of its function during cell cycle regulation. Studies have demonstrated a role for Cdh1, the mammalian homolog of Rap/Fzr, in axon growth through its interaction with the transcriptional corepressor SnoN (Konishi et al. 2004; Stegmuller et al. 2006). Furthermore, in vitro cell culture studies using neuroblastoma cell lines and silencing of Cdh1 in postmitotic cerebellar granule neurons demonstrate that the DNA-binding protein inhibitor of differentiation 2 (Id2) is a target for Cdh1-mediated ubiquitination (Lasorella et al. 2006). Our results show that Rap/Fzr is involved in glia differentiation and are consistent with other data that demonstrate that Cdh1 targets transcriptional regulators involved in the differentiation program of the developing nervous system. Thus, in addition to its role in the regulation of cell cycle progression, Rap/Fzr/Cdh1 promotes neuron formation and inhibits gliogenesis. Our studies here lend further support to the idea that ubiquitination functions as a key regulatory mechanism during nervous system development.