Drosophila gef26 is required presynaptically for normal synaptic growth
To identify genes involved in the regulation of synaptic development, we performed an anatomical screen on 1500 independent EP insertion lines [20, 21]. We inspected third instar larval NMJs using the axonal membrane marker anti-HRP. In this screen, we isolated an insertion (G3533) localized in the first intron of the Drosophila gef26 gene (CG9491). These mutants displayed NMJ overgrowth with an excessive formation of small “satellite” boutons (data not shown), which protrude from parental boutons located at primary axon terminal arbors.
To determine the null phenotype of gef26 at the NMJ, we utilized the transheterozygous combination of gef26
6, a previously reported null allele [19, 22], and the Df(2 L)BSC5 deficiency (henceforth referred to as Df) to delete the gef26 locus. A significant synaptic overgrowth phenotype was observed at every glutamatergic type-I NMJ in gef26
6
/Df third instar larvae. To quantify the gef26 phenotype, we measured overall bouton number and satellite bouton number at NMJ 6/7 and NMJ 4 from abdominal segment 2 (Fig. 1a, b; Additional file 1: Table S1). Compared with wild-type controls (w
1118), bouton number normalized to muscle surface area in gef26
6
/Df larvae was increased by 24% at NMJ 6/7 and by 51% at NMJ 4. At the same time, satellite bouton number in gef26
6
/Df was increased by 39% at NMJ 6/7 and by 219% at NMJ 4. Comparable synaptic growth defects were observed in larvae homozygous for gef26
6 (Fig. 1a, b).
To determine whether gef26 function is required pre- or postsynaptically for normal synaptic growth regulation, we expressed a gef26 cDNA transgene (UAS-gef26) in gef26
6
/Df mutants under the control of tissue-specific GAL4 drivers. Expression of UAS-gef26 using a neuronal driver (C155-GAL4) fully rescued the NMJ growth defect of gef26 mutants (Fig. 1b). In contrast, expression of UAS-gef26 in all somatic muscles using the BG57-GAL4 driver failed to rescue the NMJ growth defect (Fig. 1b), suggesting that Gef26 functions presynaptically to restrain synaptic growth at the NMJ.
Additional evidence for a presynaptic requirement for Gef26 was provided by assessment of the effect of RNA interference (RNAi)-mediated knockdown of Gef26 expression. Neuronal expression of a dsRNA-fragment of gef26 (UAS-gef26
RNAi) using C155-GAL4 increased both bouton number and satellite bouton number and mimicked the gef26 loss-of-function mutation, whereas muscular expression of the same dsRNA using BG57-GAL4 had no effect (Additional file 2: Figure S1a, b; Additional file 3: Table S2). This result supports the notion that Gef26 acts in presynaptic neurons to restrain synaptic growth at the NMJ.
We further characterized satellite boutons at gef26 mutant NMJs using several synaptic markers. Satellite boutons contained the active zone antigen NC82 and the synaptic vesicle marker cysteine-string protein (CSP) (Additional file 2: Figure S1c, d). In addition, satellite boutons were found to recruit the subsynaptic reticulum (SSR) marker discs-large (Dlg). Finally, NC82 in satellite boutons was nicely juxtaposed to the essential glutamate receptor subunit GluRIIC (Additional file 2: Figure S1e, f). Thus, satellite boutons in gef26 mutants display the anatomical hallmarks of functional synapses.
Gef26 acts through Rap1 to regulate synaptic growth
Since Gef26 acts via Rap1 to mediate various developmental processes [15,16,17, 19, 22], we decided to investigate whether Rap1 is the major target for Gef26 in the regulation of synaptic growth. We began by investigating whether loss of rap1 produces NMJ phenotypes similar to those caused by gef26 loss-of-function mutations. For this purpose, we analyzed NMJ morphology in third instar larvae homozygous for the rap1
MI11950 allele (hereafter referred to as rap1
M) harboring a Minos element within the rap1 gene. Compared with wild-type controls, both overall bouton number and satellite bouton number in rap1
M mutants were significantly increased (Fig. 2a, b; Additional file 4: Table S3). To confirm the requirement for rap1 in the proper regulation of synaptic growth, we also examined NMJ morphology in third instar larvae expressing rap1 dsRNA (UAS-rap1
RNAi) under the control of C155-GAL4. This genetic manipulation significantly increased overall bouton number and satellite bouton number (Additional file 5: Figure S2a, b; Additional file 6: Table S4). In contrast, muscular expression of UAS-rap1
RNAi did not noticeably alter NMJ morphology (Additional file 5: Figure S2a, b; Additional file 6: Table S4). Thus, loss of presynaptic rap1 produces gef26-like phenotypes at the NMJ.
Next, we assayed the transheterozygous interaction between gef26 and rap1 during synaptic growth. Heterozygous gef26
6/+ or rap1
M/+ larvae displayed normal NMJ morphology. However, overall bouton number and satellite bouton number were both significantly increased in transheterozygous gef26
6/+; rap1
M/+ larvae compared with single gef26
6/+ or rap1
M/+ heterozygotes (Fig. 2b). This type of genetic interaction suggests that Gef26 and Rap1 function in the same pathway.
Finally, we explored the epistatic relationship between gef26 and rap1. Neuronal overexpression of dominant-active Rap1-Q63E (UAS-rap1
CA) using C155-GAL4 produced an NMJ undergrowth phenotype with fewer synaptic boutons (Fig. 2b). Importantly, neuronal overexpression of UAS-rap1
CA was able to induce a similar phenotype even in the gef26
6
/Df background (Fig. 2b), indicating that the overactivity of Rap1 completely suppresses the synaptic overgrowth in gef26 mutants. These results suggest that Gef26 acts upstream of Rap1 to restrain synaptic growth at the NMJ.
Gef26 and Rap1 regulate synaptic growth via inhibition of BMP signaling
Previous studies have identified Gbb as a key retrograde signal that stimulates synaptic growth at the NMJ [4,5,6,7, 23]. Consistently, elevation of BMP signaling, which can be achieved by either presynaptic overexpression of a dominantly active Tkv receptor or loss of the inhibitory Smad Daughters against decapentaplegic (Dad), causes synaptic overgrowth with excessive satellite bouton formation [9, 10], recapitulating phenotypes exhibited by gef26 or rap1 mutants. Therefore, we wondered whether Gef26 and Rap1 might regulate synaptic growth by inhibiting BMP signaling. To test this possibility, we first examined the transheterozygous interaction between gef26 or rap1 and dad at the NMJ. Like gef26
6/+ and rap1
M/+ larvae, heterozygous dad
J1E4/+ larvae displayed normal NMJ morphology (Fig. 3a, b; Additional file 7: Table S5). In contrast, both overall bouton number and satellite bouton number were significantly increased in transheterozygous gef26
6/+; dad
J1E4/+ and rap1
M,+/+,dad
J1E4 larvae compared with wild-type controls (Fig. 3a, b), suggesting a functional link between Gef26/Rap1 and the BMP signaling pathway during synaptic growth.
We next examined whether synaptic overgrowth in gef26 or rap1 mutants depends on BMP signaling. Heterozygosity for the BMP receptor gene tkv (tkv
7/+), which had no effect on NMJ morphology in a wild-type background, suppressed synaptic overgrowth in gef26
6
/Df or rap1
M/rap1
M mutants (Fig. 3c, d; Additional file 7: Table S5). Moreover, removal of both copies of tkv (tkv
1
/tkv
7) in the gef26
6
/Df background caused a synaptic undergrowth phenotype, which was similar to that of tkv
1
/tkv
7 mutants (Fig. 3c, d). Thus, BMP signaling is necessary for synaptic overgrowth in gef26 or rap1 mutants.
Finally, we directly tested the role of Gef26/Rap1 in inhibiting BMP signaling by assaying P-Mad levels in gef26 and rap1 mutants. P-Mad accumulation at NMJ synapses and in the nuclei of ventral nerve cord (VNC) motoneurons was significantly increased in gef26
6
/Df or rap1
M/rap1
M larvae compared with wild-type controls (Fig. 3e, f). Neuronal expression of UAS-gef26 in gef26
6
/Df mutants was capable of reversing the increase of P-Mad in motoneurons (Fig. 3f), establishing the roles of Gef26 and Rap1 as negative regulators of BMP signaling. These results support a model in which Gef26 and Rap1 restrain synaptic growth by inhibiting BMP signaling.
Gef26 and Rap1 control BMP-dependent synaptic growth by regulating Drosophila fragile X mental retardation 1 (dfmr1) expression and microtubule stability
At the Drosophila NMJ, BMP signaling has been shown to repress the expression of the dfmr1 gene [9]. The dfmr1 product (dFMRP) in turn negatively regulates the expression of the microtubule-associated protein 1B (MAP1B) Futsch [24], which promotes synaptic growth by stabilizing synaptic microtubules [25]. Therefore, we hypothesized that Gef26/Rap1 might control synaptic growth by regulating microtubule stability via the dFMRP-Futsch pathway. To test the involvement of dFMRP in Gef26/Rap1-dependent regulation of synaptic growth, we first examined the transheterozygous interaction between gef26 or rap1 and dfmr1 at the NMJ. Total bouton number and satellite bouton number were significantly higher in transheterozygous gef26
6/+; dfmr1
Δ50M/+ and rap1
M, +/+,dfmr1
Δ50M larvae than in wild-type controls, although the single heterozygotes displayed normal synaptic growth (Fig. 4a, b; Additional file 8: Table S6). In a subsequent experiment, we directly tested whether loss of Gef26 or Rap1 alters dfmr1 expression. Levels of dfmr1 mRNA were significantly lower in gef26 and rap1 mutants than in wild-type controls, as demonstrated by quantitative real-time PCR (Fig. 4c). Given the roles of Gef26 and Rap1 in inhibiting BMP signaling, these results imply that Gef26/Rap1 restrains synaptic growth by relieving BMP-dependent repression of dfmr1 transcription.
Next, we investigated whether gef26 and rap1 mutants affect synaptic Futsch levels. In wild-type NMJs, Futsch was detected as a filamentous bundle occupying the center of the presynaptic terminals. However, Futsch staining was fainter or not detectable in newly formed or terminal boutons. Futsch immunoreactivity was significantly increased in gef26 and rap1 mutant axons compared with wild-type controls (Fig. 4d, e). In addition, the number of terminal boutons with Futsch immunoreactivity (visualized as a looped or punctate structure) was significantly higher in gef26
6
/Df or rap1
M/rap1
M mutants (Fig. 4d, arrowheads). These results indicate that Gef26 and Rap1 function to limit presynaptic Futsch level.
Futsch reliably labels microtubules in presynaptic motor terminals [25]. Therefore, the above results suggest the involvement of microtubule stability in Gef26/Rap1-mediated regulation of synaptic growth. To directly test this possibility, we assayed the extent of synaptic growth in gef26 and rap1 mutants fed vinblastine, a microtubule-severing drug [26]. When vinblastine was fed at a low concentration (1 μM) that did not affect synaptic growth, it completely suppressed the synaptic overgrowth phenotype of gef26
6
/Df or rap1
M/rap1
M larvae (Fig. 4f, g; Additional file 8: Table S6). These results support the idea that Gef26/Rap1 controls synaptic growth by regulating microtubule stability via the Futsch pathway.
Gef26 regulates the endocytic internalization of the BMP receptors Tkv and Wit
We next attempted to determine how Gef26 attenuates BMP signaling. Mutations disrupting endocytosis, including endophilin (endo) and dap160, increase presynaptic P-Mad levels at the NMJ along with simultaneous synaptic overgrowth and the formation of excessive satellite boutons [10, 13, 27], suggesting that endocytosis of surface BMP receptors is an important mechanism to inhibit BMP-dependent synaptic growth. Since a similar phenotype was observed in gef26 mutants, we wondered if Gef26 regulates BMP signaling through endocytosis. To test this possibility, we first investigated genetic interactions between gef26 and mutations in endocytic genes. In heterozygous gef26
6/+, endoA
Δ4/+, and dap160
Δ1/+ larvae, total bouton number and satellite bouton number were at wild-type levels (Fig. 5a, b; Additional file 9: Table S7). In sharp contrast, both parameters were significantly increased in transheterozygous gef26
6/+; endoA
Δ4/+, or gef26
6/dap160
Δ1 larvae (Fig. 5a, b), raising the possibility that Gef26 regulates BMP-dependent synaptic growth through an endocytic mechanism. It has been proposed that Dap160 interacts with the endosomal protein Nervous wreck (Nwk) to negatively regulate synaptic growth [10, 28]. However, total bouton number and satellite bouton number were normal in transheterozygous gef26
6/+; nwk
2/+ larvae (Fig. 5b), suggesting that Gef26 and Nwk regulate BMP signaling through distinct pathways.
We then examined the impact of gef26 knockdown on the endocytic internalization of BMP receptors in neuronal BG2-c2 cells. We transiently transfected a Myc-Tkv-Flag or Myc-Wit-Flag construct into control or gef26-knockdown cells (Fig. 5c) and prelabeled the cells with an anti-Myc antibody at 4 °C. We then initiated endocytosis by incubating the cells at 25 °C for 10 min and visualized the internalization of the labeled surface receptors by Myc staining. Total Myc-Tkv-Flag or Myc-Wit-Flag was also monitored by staining for the intracellular Flag-tag after cellular permeabilization. In controls cells, we observed several Myc-Tkv-Flag- or Myc-Wit-Flag-positive intracellular puncta (Fig. 5d; data not shown). Importantly, when examined in only cells with similar fluorescence intensities of Flag staining, the number of intracellular Myc-Tkv-Flag- or Myc-Wit-Flag-positive puncta per cell was dramatically reduced in gef26-knockdown cells (Fig. 5d, e), suggesting that Gef26 is required for the endocytic internalization of BMP receptors.
Next, we determined the impact of gef26 loss-of-function on the levels of surface Tkv at the NMJ. To do this, we expressed UAS-Myc-tkv in wild-type and gef26 mutants using C155-GAL4. Surface Myc-Tkv and total HRP were measured by sequential staining with anti-Myc and anti-HRP antibodies under nonpermeant and permeant conditions, respectively. The ratio of Myc-Tkv signal to HRP signal intensity was significantly increased in gef26 compared with wild-type NMJs (Fig. 5f, g). Levels of Myc-Tkv expression were not significantly different between wild-type and gef26
6/Df animals (Fig. 5h, i). These data support a role for Gef26 in endocytic internalization of the Tkv receptor.
Given the role of Gef26 in BMP receptor internalization, we examined whether synaptic vesicle endocytosis is affected in gef26 mutant NMJs. We stimulated third instar fillets with 90 mM K+ in the presence of the styryl dye FM1–43FX. During a 1-min labeling period, dye uptake into synaptic boutons was not significantly different between wild-type and gef26
6
/Df mutant animals (Additional file 10: Figure S3a, b). This result indicates that loss of Gef26 does not grossly affect endocytosis at the presynaptic terminal of the NMJ.
Gef26/Rap1 regulation of BMP signaling is essential for neuronal survival in the adult brain
Overactivation of BMP signaling in the adult Drosophila brain induces age-dependent progressive motor dysfunction and neurodegeneration [9]. Since we established the role of Gef26 in downregulating BMP signaling at the NMJ, we investigated whether gef26 knockdown induces adult phenotypes similar to elevated BMP signaling. We first assayed the locomotor performance of C155-GAL4/+; UAS-gef26
RNAi/+ flies in a geotactic climbing assay. Compared with age-matched C155-GAL4/+ controls, 20-day-old C155-GAL4/+; UAS-gef26
RNAi/+ flies displayed a significantly reduced climbing response within 30 s (C155-GAL4/+: 16.56 ± 0.57 cm, C155-GAL4/+; UAS-gef26
RNAi/+: 9.33 ± 0.11 cm, P < 0.001; Fig. 6a, b).
We then investigated whether knockdown of gef26 expression is associated with neurodegeneration by examining histological sections of adult brains. At 2 days after eclosion, C155-GAL4/+; UAS-gef26
RNAi/+ brains had normal anatomical and histological organization (Fig. 6c). However, aged C155-GAL4/+; UAS-gef26
RNAi/+ brains exhibited progressive vacuolization (Fig. 6c, arrowheads), which is a hallmark of neurodegeneration in the Drosophila brain [29]. Vacuolization progressed at a much slower rate in C155-GAL4/+ control brains (Fig. 6d). To further characterize neurodegeneration, we performed caspase-3 and TUNEL staining on 20-day-old brains. Caspase-3- or TUNEL-positive cells were detected in C155-GAL4/+; UAS-gef26
RNAi/+ brains, but not in C155-GAL4/+ control brains (Fig. 6e; Additional file 11: Figure S4a, b). In addition, TUNEL staining revealed that neuronal knockdown of gef26 induces cell death in an age-dependent, progressive manner (Additional file 11: Figure S4b). Importantly, anti-caspase-3 signals overlapped with the neuronal marker anti-Elav, but not with the glial marker anti-Repo (Fig. 6e, arrowheads). Together, these results indicate that Gef26 activity is essential for neuronal survival in the adult brain.
Finally, we investigated whether Gef26 collaborates with Rap1 and the BMP pathway to maintain normal locomotor ability and neuronal survival. To this end, we first examined transheterozygous combinations of gef26 and rap1 or dad with respect to locomotor dysfunction. At 20 days of age, transheterozygous gef26
6/+; rap1
M/+ and gef26
6/+; dad
J1E4/+ flies displayed mildly reduced climbing response compared with age-matched gef26
6/+, rap1
M/+, or dad
J1E4/+ flies (data not shown). However, these transheterozygous flies at 30 days of age exhibited severely reduced climbing ability (Additional file 11: Figure S4c, d). We also examined transheterozygous interactions between gef26 and rap1 or dad with respect to brain neurodegeneration. At 20 days of age, heterozygous gef26
6/+, rap1
M/+, or dad
J1E4/+ flies were not distinguishable from wild-type controls with respect to the total number of vacuoles (Fig. 6f). In sharp contrast, there was a significant vacuolization in the brains of transheterozygous gef26
6/+; rap1
M/+ or gef26
6/+; dad
J1E4/+ flies (Fig. 6f), supporting a functional link between Gef26, Rap1, and the BMP signaling pathway in the regulation of neuronal survival in the adult brain.