Axonal TDP-43 Drives NMJ Disruption through Inhibition of Local Protein Synthesis

Mislocalization of the predominantly nuclear RNA/DNA binding protein, TDP-43, occurs in motor neurons of ~95% of ALS patients, but the contribution of axonal TDP-43 to this fatal neurodegenerative disease is unclear. Here, we find TDP-43 accumulation in the axons of intra-muscular nerves from ALS patients, and in motor neurons and neuromuscular junctions (NMJs) of a mouse model with TDP-43 mislocalization. This leads to the formation of G3BP1-and TDP-43- positive RNA-granules in motor neuron axons, and to inhibition of local protein synthesis in axons and NMJs. Specifically, the axonal and synaptic levels of nuclear-encoded mitochondria proteins are reduced. Clearance of axonal TDP-43 restored local translation of the nuclear-encoded mitochondrial proteins and rescued TDP-43-derived axonal and NMJ toxicity. These findings suggest that targeting TDP-43 axonal gain of function may mediate a therapeutic effect in ALS.


Introduction
Amyotrophic lateral sclerosis (ALS) is a fatal adult-onset neurological disease characterized by neuromuscular junction (NMJ) disruption and motor neuron degeneration 1,2 . An important pathological hallmark in ALS patients is the mislocalization of the primarily nuclear RNA and DNA binding protein, TAR DNA-binding protein , to the cytoplasm of motor neurons [3][4][5][6] . TDP-43 is a member of the heterogeneous ribonucleoprotein binding protein (hnRNP) family, and have roles in transcription, RNA splicing, processing, and nucleocytoplasmic transport 4,[7][8][9] . Additionally, mutations in TDP-43 were identified in a subset of ALS patients 10 . Cytoplasmic accumulation of TDP-43 has been implicated in ALS via several pathways 11 , but the mechanism sensitizing motor neurons and the NMJ to TDP-43 proteinopathy remains unclear.
One key event which develops due to TDP-43 mislocalization is the formation of phase separated cytoplasmic condensates that are associated with alterations in RNA localization and translation [12][13][14] . In addition, ALS-associated mutations in TDP-43 can directly interfere with mRNA transport 15,16 . This process is associated with development of pathological RNP condensates 16 , which were shown to deregulate mRNA localization and translation 17,18 .
The formation of RNP granules 18 and mRNA transport 19 affect localized protein synthesis, an important regulator of axonal and synaptic health in different neuronal subtypes [20][21][22][23] . Several recent studies demonstrated alterations of protein synthesis in ALS models 13,24,25 . However, most of those observations were made in non-neuronal cells or within the neuronal cell body, not in the NMJ. Given that the NMJ and motor neuron axons are the first compartments to fail in ALS 26,27 , the consequences of TDP-43 mislocalization in these compartments are key to understanding disease pathology.
To study the effect of TDP-43 mislocalization on the NMJ in a precise and controlled environment, we used a neuromuscular co-culture setup in microfluidic chambers (MFCs) that we recently developed [28][29][30][31][32] . This platform can model several pathological features of ALS, such as axon degeneration, NMJ dysfunction and MN death 29,[33][34][35] . The fluidic separation between the motor neuron cell-body and axon allows the formation of functional NMJs exclusively at the distal compartment 29,[31][32][33][34][35][36] . Utilizing this system, it is possible to study and track spatiotemporal events such as localized protein synthesis at the subcellular level.
Here, we describe a novel, toxic gain-of-function of TDP-43 accumulation in axons and NMJs, which impacts synaptic local protein synthesis and provokes neurodegeneration. By employing muscle biopsies from ALS patients, we demonstrate TDP-43 pathology in distal motor neuron axons. This mislocalization leads to formation of axonal RNP-complexes that interfere with local synthesis in axons and NMJs. Furthermore, the local synthesis of nuclearencoded mitochondrial proteins is specifically inhibited, consequently promoting NMJ dysfunction. Most strikingly, the clearance of TDP-43 from axons reverses the pathological events, shedding light on the possibility for MN recovery in ALS.

TDP-43 Accumulates in ALS Patients Intra-Muscular Nerves and in Motor Axons of TDP∆NLS Mice
TDP-43 mislocalizes to the cytoplasm in ALS patient spinal cord MNs, where it is often observed as insoluble aggregate-like structures 3,4 . However, it is not clear how this process can cause distal degeneration of axons and the NMJ up to one meter away from the cell body.
To determine whether TDP-43 mis-localization also occurs in distal axons, we immuno-stained muscle biopsies from ALS patients and non-ALS controls and compared the abundance of TDP-43 in their intra-muscular nerves (Fig.1A). Axonal levels of both TDP-43 and its phosphorylated form 3 significantly increased by ~1.5-fold in ALS patients ( Fig.1B-E). Thus, TDP-43 nuclear-to-cytoplasmic mislocalization propagates to distal MN axons in ALS patients.
To study the role of TDP-43 mislocalization in ALS pathology, we utilized inducible transgenic mice expressing the human TDP-43 lacking the nuclear-localization-signal (∆NLS) through the doxycycline (dox) TET-off system. This TDP∆NLS mouse model recapitulates ALS-like MN disease pathologies [37][38][39] , including NMJ disruption, in a mechanism that is still not fully understood. To allow close monitoring of MNs in the diseased mice, TDP∆NLS mice were cross-bred to Choline-Acetyltransferase (ChAT) cre -tdTomato lox mice (hereafter ChAT tdTomato ).
Retraction of dox from adult animals, as well as in primary motor neuron culture, resulted in TDP-43 cytoplasmic mislocalization in MNs (Supp. Fig. 1A

Local Protein Synthesis.
Formation of G3BP1 positive RNP granules is strongly associated with repression of RNA translation 18 and with altered mRNA transport 17 , leading to reduced protein synthesis 18 . Axonal and synaptic protein synthesis are vital for the proper neuronal function and facilitation of synaptic transmission 22,40 . Although local protein synthesis has not yet been described in the MN synapse, the NMJ, being the largest and most distant of all synapses, most likely relies on local synthesis for its maintenance. We therefore sought to determine whether TDP-43 axonal accumulation, and the subsequent RNPs formation, impair local synthesis in MN axons and NMJs. To test this, we cultured primary MNs from TDP∆NLS in MFC, and applied O-Propargyl-Puromycin (OPP) to the axonal compartment to exclusively label newly synthesized proteins in axons. We found a substantial decrease in density of the OPP puncta in TDP∆NLS ( Fig. 3A-B) demonstrating reduced local synthesis upon cytoplasmic accumulation of TDP-43 in axons. The level of the local synthesis impairment we observed in TDP∆NLS was similar to that seen by inducing RNP-granule formation in axons with compartmental application of sodium-arsenite (NaAsO2), or by direct pharmacological inhibition of local synthesis using Anisomycin and Cycloheximide (Sup. Fig. 3A-I). Next, we aimed to evaluate local synthesis also in the most distant point of the MN axon, the NMJ. To that end, we performed co-cultures of TDP∆NLS MNs with healthy-puromycin resistant muscles (Sup. Fig. 3J-L) and quantified the OPP puncta density within in-vitro NMJs. We identified robust NMJ protein synthesis in the control co-culture system. However, the majority of TDP∆NLS NMJs were devoid of OPP signal, or had very low OPP density, indicating impaired local synthesis ( Fig. 3C-D). Furthermore, we observed a similar decrease in OPP density in NMJs following NaAsO2 application only in the NMJ side of the compartmental MFC platform (Sup. Fig. 4M-N).
Following these findings, we examined the extent of protein synthesis and its interference in adult TDP∆NLS mice. To that end, we labeled SNs and tibialis anterior (TA) muscles with OPP immediately after their dissection (Sup. Fig. 4A; see methods) 20 4B). Next, we quantified the OPP signals that co-localized with ChAT in the NMJ presynapse (Sup. Movie. 1) and identified a significant reduction in protein synthesis in NMJs from TDP∆NLS mice ( Fig. 3G-H). Taken together, our in-vitro and in-vivo evidence demonstrate that TDP-43 axonal mislocalization disrupts axonal and presynaptic local synthesis at the NMJ.

Mitochondrial Proteins and Decreases Mitochondria-Associated Local Synthesis
To determine which proteins are primarily affected by TDP-43 accumulation and local synthesis reduction in axons, we performed proteome analysis of SN axoplasm samples from TDP∆NLS and LM control mice (Fig 4A). This revealed a global reduction in nuclear-encoded mitochondrial proteins (Mitocarta 2.0 41 ), including respiratory chain complex proteins (Fig 4A-C, Sup. Table 1), implying that they are directly influenced by TDP-43 axonal accumulation and local synthesis inhibition. Additionally, the TDP∆NLS axoplasm was enriched with proteins from the "ribonucleoprotein complexes" family, including also TDP-43, thus reinforcing our observations regarding formation of RNP condensates in axons (Fig. 4C). mRNAs of nuclear-encoded mitochondrial proteins are among the most abundant mRNAs in MN axons 42,43 , therefore it is likely that the TDP-43-mediated reduction in local translation will primarily affect mitochondrial proteins. To determine whether the reduction in mitochondrial proteins occurs at the transcriptional or translational level, we performed RT-qPCR analysis on SN axoplasm and quantified the mRNA levels of three nuclear-encoded mitochondrial proteins: ATP5A1, Cox4i1, and Ndufa4, which were downregulated in TDP∆NLS axons (Protein Log2FC over LM= -1.83; -1.21; -4.17 respectively). Conversely, we demonstrated that the relative axoplasmic mRNA levels of those proteins were unchanged and even moderately increased in TDP∆NLS samples ( Fig 4D), implying that TDP-43 accumulation impairs the local translation of these mRNAs.
Localized translation in axons of some nuclear encoded mitochondrial proteins was recently shown to occur in proximity to the mitochondria 21 . To understand the relationship between TDP-43 accumulation and mitochondria-related local synthesis, we tested the extent of colocalization between mitochondria and OPP-labeled newly synthesized proteins in control versus TDP∆NLS axons. This analysis detected a profound reduction in the co-localization of mitochondria with OPP in TDP∆NLS axons ( Fig. 4E-F). Next, we determined whether mitochondrial activity is directly dependent on local translation by measuring mitochondria membrane potential using TMRE dye (an indicator for mitochondria function) after axonal application of the protein synthesis inhibitor anisomycin, as well as following formation of RNP granules via NaAsO2 treatment ( Fig 4G). Inhibition of protein synthesis using both methods reduced axonal mitochondria membrane potential ( Fig 4H). Thus, interfering with axonal local synthesis alters the function of axonal mitochondria.
Hence, we sought to identify whether the lack of mitochondria-related local synthesis due to TDP-43 axonal mislocalization also affects the function of axonal mitochondria. To this end, MN axonal mitochondria were challenged with a transient 4h protein synthesis inhibition using anisomycin, followed by a 24h recovery period. While TDP∆NLS axonal mitochondria failed to recover their initial TMRE signal, the membrane potential of mitochondria in control axons was recovered ( Fig. 4I-J). Taken together, this suggests that mitochondrial proteins are especially vulnerable to the local synthesis inhibition mediated by TDP-43 axonal accumulation.

Inhibition Leads to Neurodegeneration
Thus far, our observations demonstrated that mislocalized TDP-43 forms RNP condensates along MN axons that interfere with protein synthesis events, specifically those of mitochondrial proteins. However, the functional outcome of this abnormal process is unclear. To this end, we first tested whether NMJ activity is dependent on general mitochondrial health using mitochondrial-targeted Killer-Red fusion protein (MKR) 40 . MKR was introduced into MNs in coculture via lentiviral infection prior to muscle culturing. Upon NMJ formation, we damaged synaptic mitochondria by exclusively irradiating pre-synaptic MKR-expressing mitochondria Finally, having identified that both mitochondria and local synthesis are essential for the NMJ, we aimed to determine if MN axons exhibit neurodegeneration in response to local synthesis inhibition. An evaluation of axon health over time revealed extensive degeneration 24-hour exposure to local synthesis inhibition (Sup. Fig. 5). Importantly, TDP∆NLS axons had a significantly increased sensitivity to local synthesis inhibition, displaying ~20% more degeneration compared to control axons ( Fig. 5I-K) after 24-hours. Altogether, both mitochondrial integrity and the local synthesis that supports it play crucial roles in maintaining axonal integrity. These processes are disrupted following TDP-43 mislocalization, which could eventually lead to NMJ dysfunction and neurodegeneration.

Restoring TDP-43 Localization Recovers Local Translation of Nuclear-encoded Mitochondrial Proteins and Enables NMJ Reinnervation.
Mislocalization of TDP-43 is a pathological hallmark in the majority of ALS cases, and mutations in TDP-43 cause familial ALS, indicating a converging mechanistic role of TDP-43 dysfunction in ALS pathogenesis [3][4][5]10 . We have shown here that TDP-43 propagates also to ALS patient axons (Fig.1). To study the effect of axonal TDP-43 clearance, similar to what was previously observed in MNs cell bodies 39 , we investigated the ability of TDP∆NLS mice to recover after ceasing the expression of hTDP-43∆NLS and allowing endogenous TDP-43 to redistribute to its normal localization. We employed a recovery paradigm in which doxycycline was reintroduced into the diet of TDP∆NLS mice after initial deprivation (same was done for in vitro in primary MNs). After re-introducing doxycycline, we observed a decrease in the levels of hTDP-43 in SN axoplasm that was accompanied by a significant decrease in total TDP-43 levels ( To determine whether the clearance of TDP-43 axonal accumulation also releases the inhibitions on axonal and pre-synaptic local synthesis, we performed OPP labeling on MNs following a similar recovery paradigm. Strikingly, quantification of the OPP signal in the presynaptic side of NMJs revealed that MN protein synthesis returns to full capacity, both in-vitro Next, given the sciatic proteome analysis, we analyzed whether TDP-43 mislocalization controls the local translation of nuclear-encoded mitochondrial genes in NMJs. First, by colabeling NMJs in EDL muscles with OPP, and against the mitochondrial proteins Cox4i and ATP5A1, we showed that both proteins colocalize with OPP signals in NMJs, implicating that these undergo local translation (Fig. 6J-K, Sup. Fig. 7C-D). Importantly, quantification of the pre-synaptic occupancy of Cox4i or ATP5A1 signals (Sup. Movies 6) validated that they are deficient in NMJs of TDP∆NLS mice, and that clearance of TDP-43 restored their proper localization (Fig. 6L, Sup. Fig. 7E). Testing for the extent of colocalization between either Cox4i or ATP5A1 with pre-synaptic OPP indicated that their depletion in NMJs is dominated by reduction in their local synthesis, which was also restored upon TDP-43 clearance (Fig.   6M, Sup. Fig. 7F).
Finally, to understand the functional impact of TDP-43 mislocalization on NMJ degeneration, and whether it could be reverted by applying the recovery paradigm, we measured the innervation rate in co-cultures of TDP∆NLS MNs and healthy muscles. This demonstrated that mislocalized TDP-43 facilitates NMJ dysfunction and disruption in co-cultures ( Fig. 6N-O).
Altogether, we show that the pathological mislocalization and accumulation of TDP-43 in MN axons disrupts axonal and synaptic local synthesis. This leads to altered mitochondrial protein turnover in axons and the NMJ, and eventually sensitizes the entire synapse to degeneration ( Fig.7), a process which is reversible upon TDP-43 clearance.

Discussion
TDP-43 cytoplasmic mislocalization is a pathological hallmark of ALS, in both sporadic and familial cases [3][4][5] . Previous research on TDP-43 focused on the outcomes of cytoplasmic mislocalization in MN cell bodies 11,14 . Nonetheless, TDP-43 is regularly found in axons 44 , where it serves a role in shuttling and localization of mRNAs 15 . Recent reports also revealed that TDP-43 is important for proper axonal local synthesis 45,46 . Here, we demonstrate that ALS patients display an increase in TDP-43 levels in intramuscular nerves, suggesting a forward propagation of TDP-43 to distal axons. We validate this process using an inducible model that mimics the cytoplasmic mislocalization of TDP-43. We then show that TDP-43 axonal accumulation elicits the formation of RNA and G3BP1 containing RNP granules in MN axons that consequently interfere with axonal and pre-synaptic local synthesis. This leads to depletion of nuclear-encoded mitochondrial genes. As we show, mitochondria related local synthesis is essential to maintain the axon and the NMJ, and interference with it leads to neurodegeneration (Fig. 7). Finally, we demonstrate that inhibition of local synthesis is reversible, thus providing novel findings regarding the mechanisms by which MN can cope with temporary insult to their local synthesis capacities and to mitochondrial alterations.
The concept of local synthesis in neuronal processes has mainly been studied in regenerative contexts 18,20,23 , but recent data also suggests it is critical for understanding neurodegenerative disease mechanisms 21,24 . Our findings highlight how increased abundance of TDP-43, an RNA-binding protein hypothesized to play an important role in local synthesis, can become harmful when mobilized extensively into axons. This is associated with TDP-43 induced formation of phase separated cytoplasmic RNP accumulations 12,14,47 . Recently, G3BP1 positive RNP granules were shown to inhibit translation 18 . We show that translation inhibition is strongly implicated upon TDP-43 axonal accumulation, which is associated with G3BP1 RNP colocalization (Fig. 2). Future work will be needed to further analyze the mechanisms through which local synthesis is regulated by formation of axonal RNP condensates.
A fundamental finding in this research is that local translation in NMJs is performed to a much greater extent than in axons (Fig. 3B vs. 3D). As NMJs are enriched with mitochondria 31,48,49 , and due to mitochondrial dependency on local synthesis 21,50 (Fig 4G-H), the enhanced local synthesis in NMJs can be attributed to its mitochondrial density. This suggests that the high polarization of MNs leads to higher dependency on local synthesis of mitochondrial proteins.
As we show, the NMJ relies on mitochondria activity (Fig. 5A-D) 31 and on local synthesis (Fig.   5E-G). Therefore, interference of synaptic local synthesis might initiate local energy deficiency that ultimately leads to NMJ degeneration. Taken together, these may supply a partial explanation for how TDP-43-mediated reduction in local synthesis specifically sensitizes the NMJs to rapid degeneration, and even further, why NMJ degeneration is an early pathology in ALS 26,51 . Further research will be needed to reveal the sequence of events, focusing on the initiation of mitochondrial toxicity in the NMJ. After Dox retraction, all hTDP-43∆NLS mice were weighted weekly to track disease progression.
The colony was maintained by breeding with ICR mice (Institute of Animal Science, Harlan).
All animal experiments were approved and supervised by the Animal Ethics Committee of Tel-Aviv University.

Human muscle biopsy for intra-muscular nerve staining
Intra-muscular nerve staining was performed on muscle biopsies from 3 ALS patients and 5 non-ALS patients. All clinical and muscle biopsy materials used in this study were obtained with written informed consent during 2016-2020 for diagnostic purposes followed by research application, approved by the institutional review board. Deltoid, quadriceps or gastrocnemius skeletal muscle samples were excised via open biopsies and pathological analysis was performed at the neuromuscular pathology laboratory at Sheba Medical Center, Ramat-Gan, Israel. All 3 ALS patients were diagnosed with clinically definite or probable ALS according to El Escorial criteria 52 . Control muscles included a variation of findings, which were consistent with a diagnosis of normal muscle, severe, chronic ongoing denervation and reinnervation due to spinal stenosis, necrotic autoimmune myopathy, type 2 fiber atrophy due to disuse and overlap myositis syndrome.
Frozen muscle biopsies were than cryo-sectioned to 10µm thick slices, mounted onto slides and air dried for 30 minutes in room temperature (RT). Sections were than washed in PBS, fixed in 4% PFA for 20 min, and permeabilized with 0.1% Triton, and blocked with 5% goat serum (Jackson Laboratories) and 1 mg/mL BSA (Amresco). Sections were than incubated ON with appropriate antibodies overnight at 4°C in blocking solution [rabbit anti TDP43 or rabbit anti phospho TDP-43 (both 1:1,000, Proteintech), Chicken anti NFH (Abcam, 1:1,000).
Sections were washed again and incubated for 2 hours with secondary antibodies (1:1,000, Jackson Laboratories and ThermoFisher), washed and mounted with ProLong Gold (Life Technologies).

Microfluidic chamber preparation
Our MFCs design was recently published 53  Radial PDMS molds (Fig. 1G-J) were designed and fabricated with SU-8 photoresist protocol 54 in the Tel-Aviv University Nano and Micro Fabrication Center. as described in the following For SN RNA extraction, SN axoplasm was obtained from 2 adult mice SNs in a tube containing 100µL PBS and protease inhibitors, cut into small pieces and gently squeezed on ice. The axoplasm was then centrifuged at 10,000 G for 10min at 4°C. RNA was extracted using the RNAeasy micro kit (Qiagen) according to manufacturer's protocols.

PCR and RT-qPCR
Reverse Transcription was performed with High-capacity Reverse Transcription cDNA kit using random primers (Thermo Fisher Scientific). Standard PCR was done to test radial chambers axonal purity using KAPA ReadyMix using the following primers:

Standard PCR primers
Gene forward primer Reverse primer

GTATGGAATCCTGTGGCATC AAGCACTTGCGGTGCACGAT
qRT-PCR of SN axoplasm was done for the following genes: PolB, mitochondrial-RNR1, Cox4i, ATP5A1 and NDUFA4. Mitochondrial-RNR1 gene was used as a reference gene when calculating ∆CT, as the we aim to quantify relative mRNA levels of nuclear-encoded mitochondrial genes as a part of total axonal mitochondria.

qPCR primers Gene
Forward primer Reverse primer

Immunofluorescent staining for cryosections
SN and SC sections were prepared from fixating respective tissues in 4% PFA for 16 hours at 4°C, then incubation with 20% sucrose for 16 hours at 4°C, and cryo-embedding in Tissue- Tek OCT compound (Scigen). Tissues were then cryo-sectioned to 10µm thick slices, washed with PBS, followed by permeabilized and blocking in solution containing 10% goat serum, 1mg/mL BSA and 0.1% Triton in PBS for 1h. Later the sections were incubated ON at 4°C with primary antibody rabbit anti TDP-43 (Proteintech, 1:2,000), followed by 2h incubation at RT with secondary antibody (Jackson laboratory, 1:1,000), wash with PBS and mounting with ProLong Gold (Life Technologies) containing DAPI nuclear staining.

Whole mount NMJ staining
Gastrocnemius (GC), Tibialis Anterior (TA) or Extensor Digitorum Longus (EDL) muscles were dissected from adult mice, cleared from connective tissue and kept in 4% PFA until use.

Fluorescence microscopy and image analysis
Confocal images were captured using Nikon Ti microscope equipped with a Yokogawa CSU X-1 spinning disc and an Andor iXon897 EMCCD camera controlled by Andor IQ3 software.
Phase-contrast movies of muscle contraction were acquired using the same microscope in Epi-mode and images were captured with an Andor Neo sCMOS camera. All live imaging experiments were performed with 5% CO2 and 37°C humidified using in-situ microscope setup.

O-Propargyl-Puromycin (OPP) labeling ex-vivo
OPP was used to label protein synthesis in freshly dissected TA/EDL muscles and SNs.

LC-MS/MS analysis and Data Processing
Proteome analyses were performed using an Easy nLC 1000 ultra-high performance liquid chromatography (UHPLC) coupled to a QExactive Plus mass spectrometer (Thermo Fisher Scientific) with the same settings as described before 56  Significance cutoff was set to a log2 fold change of at least ± 0.58 and a -log10 p-value of 1.3. An ROI was then marked around the axons that overlap with the muscles, which was later frapped with a 560nm laser (100 repeats of 200µS). High-speed image sequences of calcium transients were acquired before and after 560nm laser irradiation, and the percent of change in contraction rate post/pre was calculated for each muscle.

Lentivirus production and infection
Lentivirus particles were used to infect MNs with the MKR gene. We used second generation packaging system. The helper pVSVG and pGag-Pol were gifts from Prof. Eran Bacharach

Protein synthesis inhibition functional analysis
Analysis of axon degeneration following protein synthesis inhibition was performed by culturing either WT MNs from HB9::GFP embryos, or ∆NLS and control MNs from hTDP43∆NLS embryos in the proximal compartment of MFC. Once axons have extensively crossed to the distal compartment, or after 10 days (for hTDP43∆NLS cultures), protein synthesis inhibitors were added to the distal (axonal) compartment while maintaining a higher volume of medium in the proximal compartment to prevent exposure of the cell-bodies to inhibitors. Puromycin (100µg/mL) or Anisomycin (40µM; only for HB9::GFP MN) were applied exclusively to axons in CNB medium. Images of axons were acquired at low magnification before, and after 16 (hTDP43∆NLS) and 24 hours to monitor the extent of axon degeneration.
Analysis of NMJ function following protein synthesis inhibition with puromycin was performed by co-culturing WT MNs with primary muscles transfected with empty PQCXIP-mCherry vectors expressing PAC gene for puromycin resistance. After 12 days in co-culture, once cultures matured and NMJ were formed, Puromycin (100µg/mL) was added exclusively to the distal (NMJ) compartment for 16 hours. The proximal compartment was kept with higher volume of medium for allowing puromycin to act only locally within the NMJ compartment.
After 16 hours, cultures were labeled with OGB, and the calcium activity of axons and muscles was recorded. Analysis of the percent of muscles with calcium transients in co-culture, was performed on muscles with at least one overlapping axon. Only muscles that expressed mCherry (as a reporter for the expression of PAC) were used for this analysis.

Co-culture calcium imaging
OGB lyophilized stock (Life technologies) was resuspended with 20% (w/v) Pluronic acid for a stock concentration of 3mM Stock was diluted 1:1,000 in the appropriate medium. OGB was incubated with cultures for 40 minutes in 37°C, 5% CO2 incubator, and then washed 3 times with culture medium prior imaging. Calcium transients in axons and muscles were recorded in a spinning disk confocal microscope equipped with an EMCCD camera with X40 oil objective using 488nm laser. Image sequences of 1,000 frames were acquired at frame rate of 25 FPS.
Image analysis was performed using the "Time Series Analyzer V3" plugin for FIJI. Briefly, the mean OGB values for a Region of Interest (ROI) were plotted over the complete movie length.
This assisted us to determine whether or not a certain muscle was active, and whether the activity was paired with neuronal firing. For figure labeling, axon endings on muscles, which also had high basal OGB signal were considered as NMJs.

Statistical analysis
Statistical parameters and test used are noted in Figure legends. Threshold for determining statistical significance was P < 0.05. All statistical analysis was performed with Graph-pad Prism 7.