AP2 Regulates Thickveins Trafficking to Attenuate NMJ Growth Signaling in Drosophila

Abstract Compromised endocytosis in neurons leads to synapse overgrowth and altered organization of synaptic proteins. However, the molecular players and the signaling pathways which regulate the process remain poorly understood. Here, we show that σ2-adaptin, one of the subunits of the AP2-complex, genetically interacts with Mad, Medea and Dad (components of BMP signaling) to control neuromuscular junction (NMJ) growth in Drosophila. Ultrastructural analysis of σ2-adaptin mutants show an accumulation of large vesicles and membranous structures akin to endosomes at the synapse. We found that mutations in σ2-adaptin lead to an accumulation of Tkv receptors at the presynaptic membrane. Interestingly, the level of small GTPase Rab11 was significantly reduced in the σ2-adaptin mutant synapses. However, expression of Rab11 does not restore the synaptic defects of σ2-adaptin mutations. We propose a model in which AP2 regulates Tkv internalization and endosomal recycling to control synaptic growth.

At the Drosophila NMJ, the retrograde BMP signaling is initiated by secretion of Glass bottom boat (Gbb) from the muscle and the motor neurons (Goold and Davis, 2007). Gbb binds to Wit (a Type II receptor) as well as Tkv and Sax (Type I receptors) at the presynaptic nerve terminals to control NMJ growth and function (Aberle et al., 2002;Marqués et al., 2002;McCabe et al., 2003). Gbb binding to Wit triggers the tetramerization of BMP receptors that, in turn, phosphorylates the Smad transcription factor, Mothers against decapentaplegic (Mad). Following BMP activation, these receptors are endocytosed for retrograde transport to the motor neuron nuclei in the soma (Rodal et al., 2011;Smith et al., 2012;Vanlandingham et al., 2013).
Multiple studies have shown a tight correlation between defective endocytosis, altered synapse growth, and elevated synaptic pMad levels, implicating increased BMP signaling in synaptic development (O'Connor-Giles et al., 2008;Ball et al., 2010;Nahm et al., 2010;Shi et al., 2013;Piccioli and Littleton, 2014). One such study has shown that Nwk, an F-BAR and SH3 domain-containing protein that negatively regulates synaptic growth, interacts with Tkv along with Dap160 and Dynamin (both endocytic proteins) to attenuate retrograde BMP signaling during NMJ growth (O'Connor-Giles et al., 2008). Endocytic and endosomal pathways are, therefore, critical to controlling both the activity and localization of signaling proteins that regulate synaptic growth (Rodal et al., 2011;Cosker and Segal, 2014). Defects in intracellular trafficking can also lead to enhanced signaling from the cellular compartments (like endosomes), affecting the synapse development (Di Fiore and De Camilli, 2001;Dubois et al., 2001;Sweeney and Davis, 2002;Rodal et al., 2011). In the neuronal context, the efficacy of intercellular signaling is regulated by the trafficking of activated receptor/ligand complexes following endocytosis from the presynaptic membrane.
Tightly regulated endocytic transport of BMP receptors relies on the spatiotemporal regulation of Rab GTPase function (Kelly et al., 2012). The Rab-family of GTPases regulates the progression of receptor endocytosis and participates in the successive steps of membrane maturation, receptor transport, and turnover (Horgan and McCaffrey, 2011). In particular, Rab5 regulates vesicle formation and is associated with early endosomes, while Rab7 and Rab11 associate with late and recycling endosomes, respectively (Chavrier et al., 1990;Ullrich et al., 1996). Endosomal trafficking of BMP signaling complexes at the nerve terminals is known to fine-tune the intensity and persistence of BMP signaling (Cosker and Segal, 2014;Deshpande and Rodal, 2016). Altered distribution or misregulation of Rab11 has been shown to suppress Tkv trafficking from early endosome to presynaptic membrane resulting in elevated BMP signaling (O'Connor-Giles et al., 2008;Rodal et al., 2008;Liu et al., 2014;Deshpande and Rodal, 2016).
An important, yet enigmatic question, is the correlation between defective Clathrin-mediated endocytosis (CME) and aberrant synaptic growth. For instance, it is not known whether specific modes of endocytosis externalize specific cargos whose trafficking defect perturbs synaptic signaling. CME, the major endocytic pathway, is required not only for basal synaptic transmission at nerve terminals but also for synapse development (Koh et al., 2004;Dickman et al., 2006;Choudhury et al., 2016). For instance, perturbations in CME resulting from mutations in Dynamin, AP2 subunits, Endo, or Synj all exhibit NMJ structural defects resulting in increased number but decreased size of synaptic boutons in Drosophila (Dickman et al., 2006;Choudhury et al., 2016). It is unclear whether the NMJ structural defects associated with the endocytic mutants is a consequence of deficient endosomal trafficking leading to aberrant synaptic signaling. It is likely that perturbing CME deregulates signaling modules of BMP pathway that leads to elevated pMad in the endocytic mutants (O'Connor-Giles et al., 2008;Choudhury et al., 2016).
In central synapses, AP2-dependent CME is dispensable for membrane regeneration from the presynaptic plasma membrane following high-frequency nerve stimulation (Kononenko et al., 2014). However, the critical role of CME in generating vesicles from endosome-like structures following bulk membrane endocytosis cannot be ruled out (Watanabe et al., 2014). Previous studies support a model in which compromised CME can lead to defective signalosome trafficking by trapping signaling molecules in endosomes or intermediate structures of the endosomal pathway (Dubois et al., 2001;Lloyd et al., 2002;Sweeney and Davis, 2002;Wang et al., 2007;Papagiannouli et al., 2019;Joseph et al., 2020). Another study has shown clathrin-independent role of the AP2 in the endocytic retrieval of select synaptic vesicle (SV) cargos from the presynaptic cell surface (López-Hernández et al., 2021).
Our previous study has shown elevated levels of synaptic as well as motor-nuclei pMad in s 2-adaptin mutants (Choudhury et al., 2016). In order to investigate the underlying signaling mechanisms leading to elevated pMad levels, we performed epistatic interactions between s 2-adaptin mutants with the components of BMP signaling. Introducing a mutant copy of tkv in s 2-adaptin mutants significantly reduces the NMJ overgrowth. Conversely, introducing a mutant copy of inhibitory Smad, Dad in a heterozygous s 2-adaptin background leads to NMJ overgrowth. Ultrastructural analysis of NMJ revealed accumulation of large vesicles and supports a role of s 2-adaptin in generating signalosomes containing vesicles, possibly from endosomal structures. Further analysis of vesicular trafficking using endosomal markers shows that Rab11 is reduced in s 2adaptin mutant NMJ synapses. Thus, our studies reveal a novel function of s 2-adaptin in attenuating BMP signaling by facilitating trafficking and recycling of the Tkv receptor.

Electrophysiology
All intracellular recordings were performed on wandering third instar larvae as described previously (Choudhury et al., 2016). Briefly, HL3 buffer containing 1.5 mM Ca 21 was used for larval dissection. Recordings from muscle 6 of A2 hemisegment were performed using sharp glass electrodes having a resistance of 20-25 MV. Miniature excitatory junction potentials (mEJPs) were recorded for 60 s, followed by recordings of EJPs at 1 Hz stimulation. For High-frequency recording, nerves were stimulated at 10 Hz, and EJPs were recorded for 5 min. For recording EJPs, stimulation pulse was delivered using Grass S88 stimulator (Grass Instruments, Astro-Med, Inc). The signals were amplified using Axoclamp 900A, digitized using Digidata 1440A, and acquired using pClamp10 software (Molecular Devices). Muscles with resting membrane potential between À60 and À75 mV were used for analysis. The data were analyzed using the Mini Analysis program (Synaptosoft, Decatur).

Intensity profile
Confocal images of muscle four NMJ at A2 hemisegment were used to plot the intensity profile. A single bouton section was used, and the intensities of Tkv and HRP were analyzed by drawing a line across the bouton using Fiji/ImageJ software. The graph was plotted in Microsoft Excel using the intensity values obtained from Fiji/ImageJ software. As the intensity of Tkv in control was too low to plot the graph, all the intensity values were multiplied twofold.

Electron microscopy
TEM was performed as described previously (Deivasigamani et al., 2015). Briefly, third instar larvae were dissected in cold PBS. The larval fillets were then fixed in 0.12 M cacodylate buffer containing 2% glutaraldehyde for 10 min at room temperature, transferred to a fresh fixative, and kept overnight at 4°C. The fillets were postfixed for 1 h with 2% osmium tetroxide (OsO 4 ) solution prepared in 0.12 M cacodylate buffer. The samples were rinsed with 0.12 M cacodylate buffer followed by washes with distilled water to avoid precipitation of cacodylate with Uranyl acetate. Subsequently, the samples were subjected to en bloc staining with 2% uranyl acetate. The stained fillets were again washed with distilled water and dehydrated using graded solutions of ethanol before final infiltration of the samples through propylene oxide for 30 min. Stained and dehydrated fillets were embedded in epoxy resin and hardened overnight at 60°C. Muscles embedded in epoxy resin were sectioned at 60 nm. Ultrathin sections of the muscles stained with 2% uranyl acetate (in 70% ethanol) and 1% aqueous lead citrate were examined at 120 KV on a Tecnai G2 Spirit BioTWIN (FEI) electron microscope. The number of synaptic vesicles per bouton were counted manually using the Multi-point tool in ImageJ/Fiji software and then divided by their respective bouton areas to obtain the vesicle density/mm 2 area of a bouton. For vesicle size, diameters of at least 100 vesicles from 10 bouton sections of each genotype were used for quantification. For calculating the number of synaptic vesicles docked at the active zones, only those vesicles that were touching the T-bar platform were counted. For calculating the SSR thickness, the scale bar in the images was first calibrated to the number of pixels using the Set scale function in ImageJ/Fiji software. This was followed by using the Straight-line tool to draw a line across the SSR and Measure tool to calculate the thickness.

Western blot analysis
The western blot analysis was done as previously described previouusly (Choudhury et al., 2016). Briefly, VNC from wandering third instar larvae were dissected out in ice-cold HL3 buffer and homogenized in buffer containing 50 mM Tris-HCl, pH 6.8, 25 mM KCl, 2 mM EDTA, 0.3 M sucrose, and 2% SDS in water. The homogenized sample was then mixed with an equal volume of 2Â Laemmli buffer. The protein equivalent to 50 mg was separated on 12% SDS-PAGE and transferred to Hybond-LFP PVDF membrane (GE Healthcare, GE Healthcare Life Sciences). The membrane was then blocked with 5% skimmed milk for 1 h, followed by overnight incubation with anti-Rab11 (1:2000) and anti-Tubulin (1:5000) antibody. IRDye 800 (1:10 000) was used as a secondary antibody, and signals were visualized on LI-COR Odyssey platform. The density of Western bands was quantified using ImageJ/ Fiji software.

Quantification and statistical analysis
For fluorescence quantification, images were captured using a laser scanning confocal microscope (LSM780; Carl Zeiss or FV3000, Olympus). All the control and experimental fillets were processed in the same way, and the fluorescence images were captured under the same settings for every experimental set. For bouton quantification, CSP-labeled structures were counted at muscle 6/7 of A2 hemisegment. The number of boutons from each NMJ was normalized to the respective muscle area. To calculate the bouton number, NMJs from A2 hemisegment were captured using a plan apochromat 40Â objective, 1.4 NA and all the CSP positive boutons were counted manually in ImageJ/Fiji software. For muscle area quantification, images from A2 hemisegment were captured using 20Â objective, and the area was quantified using ZEN2 software (Carl Zeiss, Germany). For bouton number quantification, the total number of boutons per NMJ was divided by the respective muscle areas. For fluorescence intensity quantification, NMJs from muscle four were captured using a plan apochromat 60Â, 1.4 NA objective. For each NMJ, the fluorescence intensity from each bouton was subtracted from the background intensity, and the average intensity was normalized to the control. The fluorescence intensity was calculated using ImageJ/Fiji software. For bouton area quantification, NMJs from muscle 6/7 at A2 hemisegment were captured, and the area was calculated by drawing a free-hand sketch around CSP positive bouton using ImageJ/Fiji software. For multiple comparisons, one-way ANOVA followed by post hoc Tukey's test, or Student's t test was used. GraphPad Prism 8 was used to plot the graph. Error bars in all the histograms represent SEM. *p , 0.5, **p , 0.01, ***p , 0.001.
Since the ultrastructural analysis of s 2-adaptin mutants revealed an accumulation of large endosome-like membranous structures similar to mutants with perturbed endocytosis or endosomal trafficking such as Rab5, Rab8, and Rab11 mutants (Shimizu et al., 2003;West et al., 2015;Inoshita et al., 2017), we hypothesized that s 2adaptin could be involved either in endocytosis of BMP receptors from the presynaptic membrane or in the endosome-dependent trafficking of the receptors. To test this possibility, we first assessed the level of Tkv receptors at the larval NMJ. Since specific antibodies against Tkv receptors are not available, we expressed an EGFP-tagged Tkv receptor transgene in the motor neurons of s 2-adaptin mutants. Interestingly, we found a significant accumulation of Tkv receptors at the mutant synapses (D42-Gal4, AP2s KG02457 /AP2s ang7 , UAS-tkv-EGFP: 427.5 6 20.20, p 0.001) when compared with control (D42-Gal4/UAStkv-EGFP: 100 6 8.49; Fig. 5A-D,I). In order to analyze the subcellular accumulation, we plotted the intensity profiles of Tkv-EGFP and HRP (a presynaptic membrane maker) in single sections of the acquired images. While in control synapses, Tkv localized both at the presynaptic membrane as well as within the bouton; we found a higher intensity peak of Tkv-EGFP at the presynaptic membranes of s 2-adaptin mutants (Fig. 5J-O). Consistent with these observations, we found that knocking down a-adaptin (D42-Gal4.a-adaptin dsRNA; 276.1 6 14.75, p 0.001)] or clathrin light chain (Clc; D42-Gal4.Clc dsRNA; 408.61 6 21.17, p 0.001) in motor neurons showed significantly increased synaptic Tkv levels ( Fig. 5E-I) with intensity profiles similar to that of s 2-adaptin mutants (Fig. 5P-U). In contrast, mutants defective in proteins involved in later stages of endocytosis did not show Tkv accumulation at the presynaptic membrane (Fig.  5I), which is consistent with previous studies (O 'Connor-Giles et al., 2008;G Zhao et al., 2015). These data indicate that endocytosis/trafficking of Tkv receptors in s 2-adaptin mutants is severely compromised, leading to their accumulation at the synaptic membranes.
In conclusion, s 2-adaptin and Rab11 increase Tkv accumulation and BMP signaling to generate similar NMJ phenotypes via independent pathways.

Discussion
Compromised endocytosis not only perturbs synaptic transmission but also has been implicated in deregulating synaptic growth as demonstrated in Endo, Synj, nwk, shi, Clc, brat, and s 2-adaptin mutant NMJs (Rikhy et al., 2002;Verstreken et al., 2002;Verstreken et al., 2003;Choudhury et al., 2016). The underlying molecular mechanism by which these proteins regulate synaptic growth, however, has only been demonstrated for nwk and brat (O' Connor-Giles et al., 2008;Shi et al., 2013). Our previous study on s 2-adaptin mutants showed no change in levels of endocytic proteins like Endo, Synj, and Dyn (Choudhury et al., 2016), prompting us to investigate the role of s 2-adaptin in synaptic growth signaling. Mutations that affect endocytosis, in general, show synaptic overgrowth and increased BMP effector, pMad. The signaling output of growth-regulating pathwaysis often dependent on intracellular traffic that in part is dependent on endocytosis of activated receptors, ultimately impinging on the BMP, JNK, or Wingless pathways (Shi et al., 2013;Deshpande and Rodal, 2016). Here, we show for the first time that s 2-adaptin genetically interacts with BMP signaling pathway at the synapse. Loss of s 2-adaptin leads to accumulation of Tkv receptor at the NMJ. Additionally, we provide evidence that s 2-adaptin regulates the localization of recycling endosomal protein Rab11.

r2-adaptin genetically interacts with BMP pathway to regulate neuronal BMP signaling
Endosomal trafficking of BMP receptors is a crucial regulatory mechanism that controls synaptic growth (Rodal et al., 2011). Various proteins interact with BMP receptors to facilitate or attenuate the BMP-dependent signaling cascade (McCabe et al., 2004;XW Zhang et al., 2017). Endocytic proteins appear to be fascinating candidates as BMP receptor interactors. Drosophila loss-of-function endocytic mutants correlate with elevated BMP signaling and neuronal overgrowth phenotype (Dickman et al., 2006;Rodal et al., 2011;Deshpande and Rodal, 2016). Consistent with this, we show that increased Tkv levels at the NMJ result in elevated BMP signaling in s 2-adaptin mutants. If the BMP pathway is responsible for the synaptic overgrowth in s 2-adaptin mutants, we reasoned that reducing the levels of BMP signaling components should rescue the NMJ phenotype. In agreement with this, we found that partially reducing BMP receptors Tkv, Wit, and cytosolic co-Smad molecule, Medea, significantly rescues the NMJ defects in s 2-adaptin mutants. Our data reveal that s 2-adaptin genetically interacts with the negative regulator of BMP signaling, the inhibitory Smad, Dad. Transheterozygotes of Dad and s 2-adaptin mutants have increased number of boutons compared with heterozygotes of either mutant alone. Consistent with this inference, neuronal expression of UAS-Dad in s 2adaptin mutant background significantly reduces the synaptic overgrowth phenotype. However, reducing BMP signaling only partially reduces the bouton size in s 2-adaptin mutants. There could be at least two plausible explanations for the partial rescue of the bouton size: first, removing only one copy of tkv may not be sufficient for rescuing this defect. Since mutating both copies of tkv results in embryonic lethality, we could not test the epistatic interactions by removing both the copies of Tkv in s 2adaptin mutants. Second, s 2-adaptin may regulate NMJ bouton size through a different signaling pathway, which continued a-adaptin dsRNA (P, Q), and D42-Gal4, tkv-EGFP/Clc dsRNA (S, T). Note that the intensity profile plot across bouton (shown in L, O, R, U as a thin line) shows that s 2-adaptin mutant (O), D42-Gal4.a-adaptin dsRNA (R), and D42-Gal4.Clc dsRNA (U) has more Tkv receptors at the membrane compared with control (L). Scale bar in T represents 2.5 mm. Figure 6. s 2-adaptin mutant synapses show a reduction in the recycling endosome marker, Rab11. A-D, Confocal images of NMJ synapses at muscle 4 of A2 hemisegment in control (A, B), AP2s KG02457 /AP2s ang7 (C, D), double immunolabeled with early endosomal marker Rab5 (represented in grayscale/green) and neuronal membrane marker, HRP (magenta). E, Histogram showing the Rab5 level in control (100.0 6 3.0) and AP2s KG02457 /AP2s ang7 (97.25 6 4.81) synapse. Error bar represents SEM; statistical analysis was done using Student's t test. ns, not significant. F-I, Confocal images of NMJ synapses at muscle 4 of A2 hemisegment in control (F, G), AP2s KG02457 /AP2s ang7 (H, I), double immunolabeled with late endosomal marker, Rab7 (represented in grayscale/green), and neuronal membrane marker, HRP (magenta). J, Histogram showing the Rab7 level in control (100.0 6 5.71) and AP2s KG02457 / AP2s ang7 (115.6 6 8.17) synapse. Error bar represents SEM; statistical analysis was done using Student's t test. ns, not significant. K-R, Confocal images of NMJ synapses at muscle 4 of A2 hemisegment in control (K, L), AP2s KG02457 /AP2s ang7 (M, N), D42-Gal4, AP2s KG02457 /UAS-AP2s , AP2s ang7 (O, P), and Rab11 ex2/93 Bi (Q, R) double immunolabeled with recycling endosomal marker, Rab11 (represented in grayscale/green), and neuronal membrane marker, HRP (magenta). Scale bar in R represents 3 mm. S, Histogram showing relative Rab11 level normalized to HRP in control (100 6 5.73), AP2s KG02457 /AP2s ang7 (61.81 6 7.11); D42-Gal4, AP2s KG02457 /UAS-AP2s , AP2s ang7 (112.9 6 8.29) and Rab11 ex2/93 Bi (14.21 6 1.57) synapses. Error bars represent SEM; statistical analysis is based on one-way ANOVA followed by post hoc Tukey's multiple-comparison test. ***p , 0.001; ns, not significant. The data supporting that the levels of Rab11 is not altered in s 2-adaptin mutants is provided in Extended Data Figure 6-1. remains to be elucidated. Overall, our data suggest that s 2-adaptin negatively regulates the BMP growth signaling pathway to attenuate synaptic growth.
r2-adaptin regulates trafficking of Thickveins at the NMJ BMP signaling has been extensively studied in the context of neuronal growth in which the activated Tkv is endocytosed and fuse with the early endosomes, where it activates downstream signaling molecules. The signaling is attenuated when these activated receptor-containing vesicles recycle back to the plasma membrane or fuse with lysosomes for degradation (Rodal et al., 2011;Smith et al., 2012). Trafficking of these receptors into and out of such endosomes provides an additional tier for spatial and temporal modulation of signal transduction. The members of the Rab family of small GTPases regulate various stages of endocytosis (Kelly et al., 2012). Our immunocytochemistry data show elevated Tkv receptor levels at the synapses and motor neuron soma (data not shown) of s 2-adaptin mutants. Besides, levels of Rab11 (known for its role in the recycling of Tkv receptor) are reduced by half in s 2-adaptin mutant synapses.
Interestingly, levels of early and late endosomes marked with Rab5 and Rab7, respectively, remain unaffected at the s 2-adaptin mutant synapses. The intensity profile of Tkv and HRP across the bouton shows that s 2adaptin mutant has a higher intensity of Tkv at the membrane, indicating that a significant proportion of the Tkv receptors are accumulated at the presynaptic membrane. The pattern of Tkv enrichment in s 2-adaptin and Rab11 mutants were distinct. While s 2-adaptin mutant synapses showed Tkv enriched at synaptic membranes, Rab11 mutants had a rather punctate distribution within the bouton. The differential distribution of these proteins could be because of their distinct roles in the neurons. While s 2-adaptin plays a critical role in retrieving the SVmembrane from the presynaptic membrane, Rab11 is a component of the endolysosomal machinery. A clear separation of Tkv signals/pixels distinguishing plasma membrane Tkv from endosomal Tkv is challenging at the NMJ, given the resolution limit of the confocal system. The difference in Tkv distribution in s 2-adaptin and Rab11 mutants also argues for additional pathways, other than Rab11, through which s 2-adaptin may regulate BMP signaling.
Functional and morphologic aspects of r2-adaptinmediated BMP signaling can be discriminated Morphologic features of synapses often dictate functional outcomes, and physiological analyses of BMP signaling mutants reveal the same. In wit mutants, the size of the NMJ is significantly reduced with concomitant reduced evoked excitatory potentials (Aberle et al., 2002;Marqués et al., 2002). Analyses of tkv and sax, co-Smad Medea, and transcription factor Mad showed that these NMJs have smaller synapses with severe functional deficits (McCabe et al., 2004). Consistent with the role of BMP-signaling in regulating NMJ growth, gbb mutant larvae also exhibit shorter NMJs with severely reduced evoked potentials (McCabe et al., 2003). s 2-adaptin mutant synapses show a modest reduction in evoked potentials, and the protein is dispensable for maintaining basal synaptic transmission (Choudhury et al., 2016). However, a rundown of EJP amplitudes during high-frequency stimulation in synaptic mutants implicated in CME, such as Endo, Synj, and Dap160, show a rapid stimulus-dependent decline in EJP amplitude that recovers following a period of rest after the high-frequency stimulation paradigm (Verstreken et al., 2003;Koh et al., 2004). In our previous study, we reported that s 2-adaptin mutants do not recover from synaptic depression even after the 90s rest after the cessation of high-frequency stimulation (Choudhury et al., 2016). This observation suggested that in addition to its requirement in synaptic membrane retrieval, the s 2adaptin function is also required during the much slower process of SV trafficking, possibly at one of the rate-limiting steps in SV regeneration. This conclusion is supported by our EM data that shows the accumulation of endosome-like structures more frequently at the mutant synapses when compared with w 1118 controls. EJP and high-frequency recordings from s 2-adaptin mutant synapses with one copy of tkv 7 did not show any rescue in synaptic function. These data drive the conclusion that partial reduction of BMP pathway components can only rescue morphologic defects in s 2-adaptin mutants but not functional aspects and that morphologic and functional deficits can be discriminated in these mutants. Besides, the partial rescue of bouton size and bouton Retrograde NMJ growth signaling in Drosophila involves Gbb ligand that is secreted from postsynaptic muscles and binds to BMP receptors on the presynaptic membrane to activate them (Aberle et al., 2002;McCabe et al., 2003). Activated receptors are then internalized through CME and fuse with early endosomes to trigger the downstream signaling cascade (Hartung et al., 2006). From early endosomes, receptors either get sorted to the Rab11-positive recycling endosomes that recycle them back to the presynaptic membrane or are sorted for lysosomal degradation (Liu et al., 2014;Deshpande and Rodal, 2016). Depletion of s 2-adaptin/AP2-complex leads to enrichment of Tkv receptors at the presynaptic membrane and/or in the early endosomes leading to elevated BMP signaling resulting in synaptic overgrowth. It is also likely that reduced synaptic Rab11 levels in s 2-adaptin might perturb Rab11-mediated recycling of Tkv leading to enhanced BMP-signaling. clustering at the NMJ argues for possible deregulation of multiple signaling pathways in s 2-adaptin mutants that remain to be explored.
Our study uncovers and extends the existing knowledge of synaptic growth signaling and endocytosis. We provide four lines of evidence on the critical role of s 2adaptin in modulating BMP-dependent synaptic growth signaling at the Drosophila NMJ. First, we show using genetics that the morphologic defects in s 2-adaptin mutant synapses can be partially rescued by introducing a mutant copy of the BMP receptors, tkv, and wit. We also show a direct epistatic interaction between s 2-adaptin and the inhibitory Smad, Dad. Second, using immunohistochemistry, we show that s 2-adaptin mutant synapses accumulate Tkv at the plasma membrane and some of these receptors that are endocytosed and make it to the early endosomes fail to recycle back to the plasma membrane. Third, our electrophysiology data establish that morphologic and functional defects can be discriminated in s 2-adaptin mutants. Finally, our electron micrographs provide evidence for the presence of large endosomes and support our conclusion that s 2-adaptin is critically required at a later step of vesicle regeneration following endocytosis from the plasma membrane.
While this study does not report a direct biochemical or epistatic interaction between s 2-adaptin and Rab11, we observed a significant reduction in Rab11 immunoreactivity in s 2-adaptin mutant synapses that could be restored by neuronal expression of s 2-adaptin transgenes. However, the structural and functional defects of s 2-adaptin mutants could not be restored by neuronal expression of Rab11 WT or Rab11 CA . This rules out the notion that s 2-adaptin phenotypes result from the observed reduction of Rab11. The Rab11 mutant used in this study is a hypomorph with substantial Rab11 protein being detected by us and others (Khodosh et al., 2006). This precludes further epistatic analyses of this mutant. The fact that Rab11 protein could still be detected in these Rab11 hypomorphic mutants may explain why pMad levels were not as upregulated when compared with s 2-adaptin mutants. Finally, the separation of Tkv immunoreactivity from the synaptic membrane versus that of the endosomal membrane is challenging, given the resolution of the imaging system used here. Our electron micrographs, however, convincingly show an accumulation of endosome-like structures close to the plasma membrane in s 2-adaptin mutant boutons. The endosomelike structures observed in the s 2-adaptin mutants are strikingly similar to structures previously reported for clathrin (Khodosh et al., 2006), AP180 (B Zhang et al., 1998), and Rab11 (Inoshita et al., 2017), among others. Thus, Tkv receptors could likely be enriched in these endosome-like compartments.
We propose a model in which s 2-adaptin/AP2-complex is required for the attenuation of BMP signaling at the Drosophila NMJ. In the absence of AP2, recycling of Tkv is compromised, which results in its enrichment at the presynaptic membrane and/or in the early endosomes leading to elevated BMP signaling and synaptic overgrowth (Fig. 9). This study thus opens new avenues where the role of other CME components and their interaction with various growth signaling pathways can be studied. Since receptor localization and regulation appear to be the central theme in modulating BMP signaling and synapse growth, it will be interesting to perform structure-function analyses of BMP receptors and identify key residues/motifs that interact with AP2 and facilitate its endocytosis. Mutating tyrosine-based signal (YXXf ) and dileucine-based signal ([DE]XXXL[LI]) motifs in Tkv and Wit could lead to further understanding of these intricate interactions.