Kinesin-3 mediated delivery of presynaptic neurexin stabilizes growing dendritic spines and postsynaptic components in vivo

A high degree of cell and circuit-specific regulation has complicated efforts to precisely define roles for synaptic adhesion proteins in establishing circuit connectivity. Here, we take advantage of the strengths of C. elegans for cell-specific analyses to investigate molecular coordination of pre- and postsynaptic development. We show that developing dendritic spines emerge from the dendrites of wild type GABAergic motor neurons following the localization of active zone proteins and the formation of immature synaptic vesicle assemblies in presynaptic terminals. Similarly, clusters of postsynaptic receptors and F-actin are visible in GABAergic dendrites prior to spine outgrowth. Surprisingly, these developmental processes occur without a requirement for synaptic activity. Likewise, the initial stages of spine outgrowth and receptor clustering are not altered by deletion of the C. elegans ortholog of the transsynaptic adhesion protein, neurexin/NRX-1. Over time, however, dendritic spines and postsynaptic receptor clusters are destabilized in the absence of presynaptic NRX-1/neurexin and collapse prior to adulthood. The kinesin-3 family member, UNC-104, delivers NRX-1 to presynaptic terminals and ongoing UNC-104 delivery is required into adulthood for the maintenance of postsynaptic structure. Our findings provide novel insights into the temporal order of synapse formation events in vivo and demonstrate a requirement for transsynaptic adhesion in stabilizing mature circuit connectivity. Contributions Summary D.O. generated strains, transgenic lines, molecular constructs, confocal and electron microscopy images, and data analysis. S.R. performed all calcium imaging experiments and analysis. C.M.L. generated destination vectors and nrx-1 constructs. D.H.H. and K.C.Q.N. aided in electron microscopy experiments and analysis. M.M.F and D.O designed and interpreted results of all experiments and wrote the manuscript.


Introduction
The capabilities of neural circuits to perform specific functions arise from the patterns of synaptic connections between their partnering neurons. The organization of these connections is circuit-specific and established through a complex process that involves the coordinated assembly and maturation of specialized pre-and postsynaptic structures on appropriate partnering neurons, and their maintenance in the mature nervous system. We now have a generalized understanding of synapse structure. Active zone (AZ) proteins and neurotransmitter-filled synaptic vesicles position near the presynaptic membrane for rapid release while neurotransmitter receptors are clustered at high density in apposition to these sites in order to ensure the fidelity of synaptic communication. Genetic studies have identifed numerous mutations that alter circuit connectivity or affect the overall structural organization of synapses (Hendi et al., 2019) (Zhen & Jin, 1999) (Richmond et al., 1999). However, for many of the synapse-associated proteins affected by these mutations, we do not yet have a mechanistic understanding of their roles in establishing synaptic connections, or how their disruption may lead to alterations in synapse development and structure. Gaining an enhanced understanding of the sequence of events involved in synapse assembly, maturation, and maintenance, and their relative timing in vivo is critical for addressing these questions.
Several prior studies have examined molecular events during synaptogenesis. In vivo studies in C. elegans and Drosophila have largely focused on the formation of the presynaptic active zone. Collectively, these studies provide compelling evidence for a model where active zone assembly occurs sequentially. While there is some variability across synapse type, the early stages of this process are generally organized by the highly conserved synaptic scaffolds SYD-2/Liprin-a and SYD-1/Rho GTPase, which then recruit additional key conserved AZ proteins such as ELKS-1/Bruchpilot, Piccolo family members, and UNC-10/RIM for subsequent stages of assembly, including clustering of Ca 2+ channels and the recruitment of synaptic nervous system enable complex, cell-type specific functions for neurexins at synapses (Missler & Sudhof, 1998) (Sudhof, 2017). For example, conditional deletion of mouse neurexins revealed strikingly different functions across synapses formed by cortical inhibitory interneurons and those formed by cerebellar climbing fibers onto Purkinje neurons (L. Y. Chen et al., 2017).
Genetic tools that enable cell-type specific analysis are therefore critical for uncovering precise functional roles of synaptic adhesion proteins within the context of individual neural circuits.
Importantly, molecular mechanisms for their trafficking and delivery to synapses also remain incompletely defined.
We previously identified finger-like dendritic protrusions from the dendritic processes of C. elegans DD GABAergic motor neurons ( Figure 1A) (Philbrook et al., 2018). Characterization of these structures by our laboratory and others pointed towards the idea that they receive synaptic input from presynaptic cholinergic motor neurons and serve analogous roles to dendritic spines in the mammalian brain (J. G. White et al., 1976) (Cuentas-Condori et al., 2019). Deletion of nrx-1, the sole C. elegans ortholog of neurexin, disrupts these dendritic spines and impairs proper localization of cholinergic receptor clusters to postsynaptic sites on GABAergic dendrites (Philbrook et al., 2018). Here we define the order and timing of pre-versus postsynaptic development at spine-associated synapses and elucidate the role of neurexin in the coordination of these processes in vivo. We find that clusters of both presynaptic proteins and postsynaptic receptors are clearly visible prior to spine outgrowth. In the absence of neurexin, immature spines and receptor clusters form initially but then disappear, indicating that neurexin is required for synapse stabilization and maturation rather than synaptogenesis. These maintenance and maturation processes are supported by kinesin-3/UNC-104 motor dependent transport of NRX-1 to presynaptic terminals. Together, our results suggest that early and ongoing axonal delivery of neurexin to sites of synapse formation is critical for stabilization of postsynaptic structures. Our findings cleanly dissect a key role for neurexin mediated adhesion

Results
We initially explored morphological features of mature spines using 3D rendering of spines from confocal imaging. We identified several morphological classes ( Figure 1B) based on the number of spine heads and the ratio between the length of the spine to the width of the spine head and spine neck. We found that a majority of DD spines can be categorized into four morphological classes (percentage ± SD): mushroom (5 ± 1%), branched (5 ± 1%), stubby (73 ± 3%), and thin (15 ± 1%). The morphological diversity we observed by light microscopy was also evident in ultrastructural studies using serial electron microscopy ( Figure S1.1). Clusters of small synaptic vesicles were also evident in neurites (of cholinergic VA/VB neurons, presumably) immediately adjacent to the tips of DD spines, indicating putative sites of synaptic input.
To explore how the neuronal cytoskeleton may contribute to spine morphology, we examined the localization of F-actin and tubulin in DD dendrites using DD neuron-specific expression GFP::UtrCH (GFP fused to the Utrophin calponin homology domain) (Chia et al., 2014) and TBA-1::GFP (GFP-tagged a-tubulin) (Yan et al., 2013) respectively. We found that the two markers occupy distinct territories. Coexpression studies showed that F-actin is strongly enriched within dendritic spines ( Figure 1C-E), while tubulin occupies the dendritic shaft ( Figure 1F,G) with more variable localization to the spine base ( Figure S1.2). Previous findings for vertebrate neurons indicate a similar segregation of actin and tubulin to the spines and shaft of the dendrite respectively (Gu et al., 2008) (Hu et al., 2008) (Jaworski et al., 2009. DD spines are positioned immediately adjacent to the axon terminals of cholinergic neurons that are predicted to be their presynaptic partners by electron microscopy ( Figure   S1.1) (J. G. White et al., 1976) (J. G. White et al., 1986) (Cuentas-Condori et al., 2019). We previously showed stimulation of cholinergic motor neurons elicited Ca 2+ responses in was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021 GABAergic motor neurons (Philbrook et al., 2018). To confirm that spines on DD GABAergic motor neurons represent sites of functional synaptic contacts, we asked whether stimulation of presynaptic cholinergic neurons was sufficient to elicit calcium responses in DD spines. We performed in vivo calcium imaging of evoked responses in spines using combined expression of a membrane-associated GCaMP6f calcium sensor in DD GABAergic neurons and a red shifted channelrhodopsin, Chrimson, in cholinergic neurons (Philbrook et al., 2018) (Figure 1H, I, J) ( Figure S1.3). Following 5 seconds of baseline recording (488 nm, 100 ms exposure), we measured calcium responses to presynaptic depolarization (5 s, 625 nm, 30 mW/cm 2 ) ( Figure   1I). We noted significant fluorescence increases in spines that were timed with the onset of stimulation and were not present in the absence of retinal. These typically occurred simultaneously across multiple spines and returned to baseline within 5 s following stimulation.
Importantly, mutation of the cholinergic vesicular transporter/VAChT unc-17, reduced evoked calcium responses by 81%, indicating the Ca 2+ responses we observed in dendrites were dependent on presynaptic acetylcholine release ( Figure 1H, I, J). Our findings are consistent with those of another recent study (Cuentas-Condori et al., 2019), and indicate that spines on DD GABAergic neurons are primary sites of synaptic contact.

The development of synapses at DD GABAergic spines
Using the synapses at DD spines as a model, we next investigated the relative timing of pre-and postsynaptic events during synapse formation. C. elegans progress through four larval stages of development (L1-L4) prior to adulthood. DD neurons undergo a well-characterized program of synaptic remodeling such that the mature circuit organization is established during the transition from L1 to L2 stage (J. G. White et al., 1978). During this period of remodeling, newly born cholinergic neurons form new synaptic connections with the ventral dendrites of DD GABAergic neurons. This synaptic remodeling event offers a well-defined temporal window to investigate de novo formation of neuron-neuron synaptic connections in vivo. We focused on understanding the time course of spine formation relative to 2 key events in synaptogenesis: 1) the formation of presynaptic release sites, and 2) the clustering of postsynaptic receptors and Factin ( Figure 2).
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021 Overall, we found CLA-1 is more discretely localized to putative synapses than SNB-1, both at earlier timepoints and throughout development, consistent with its specific localization to active zones. Immediately prior to remodeling (~12 hours after hatch), we did not observe significant localization of either presynaptic CLA-1 or SNB-1 adjacent to the DD process. Shortly after (16 hours after hatch), we noted the initial appearance of SNB-1-and CLA-1-associated fluorescence in presynaptic cholinergic processes. Over the next 4 hours (20 hours after hatch) individual CLA-1 clusters were more clearly distinguishable, suggesting CLA-1 association with developing active zone structures (Figure 2A, D, L). Synaptic vesicle fluorescence was initially diffuse at 16 hours after hatch and became more clearly organized into discrete puncta over a similar time course to CLA-1 (Figure 2A, E, L). We did not observe the emergence of DD dendritic spines until 24-32 hours after hatch, well after initial active zone formation and recruitment of synaptic vesicles ( Figure S2A, C-E). Spines continued to mature through L4 stage, increasing in both length and number (Figure 2.2C).
We next examined the clustering of postsynaptic acetylcholine receptors in DD dendrites (ACR-12::GFP). Prior to remodeling (12 hours after hatch), we did not observe detectable levels of cholinergic receptors in the DD process. Surprisingly, we noted immature receptor clusters faintly visible in the dendritic shaft by 16 hours after hatch, well prior to the emergence of dendritic spines ( Figure 2B, F, L). These receptor clusters increased in number and redistributed towards the tips of growing spines by 32 hr ( Figure 2B, F, L). We found a similar developmental trend using DD neuron-specific labeling of the LEV-10 transmembrane auxiliary protein (Gally et al., 2004), previously shown to concentrate in spines (He et al., 2019) ( Figure   S2.3). By L4 stage, receptor and LEV-10 clusters are clearly visible at the tips of mature spines.
Together, our analysis indicates that the initial stages of development of both pre-(CLA-1 and SNB-1 clusters) and postsynaptic (AChR and LEV-10 clusters) structures occur prior to spine outgrowth, raising questions about how these initial processes may be regulated.
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021 To begin to address this question, we analyzed the distribution of F-actin in dendrites of the DD neurons during synapse formation ( Figure 2B, G-L). We noted that clusters of F-actin were evident in the ventral DD process prior to the completion of synaptic remodeling (12 hours after hatch), prior to presynaptic CLA-1 and synaptic vesicle accumulation, and before the clustering of postsynaptic receptors. Dendritic F-actin-based structures became more abundant coincident with increases in the number of presynaptic CLA-1 and postsynaptic receptor clusters. To investigate this process in real-time, we used live imaging to examine the dynamics of postsynaptic F-actin (GFP::UtrCH) in the developing DD dendrite. We found that F-actin is highly dynamic during early developmental stages (16-20 hours after hatch) compared to L4 stage where the circuit has completed maturation ( Figure S2.4, Video 1, and Video 2). In young animals, GFP::UtrCH clusters often shuttled out of the cell body to the main dendritic process, perhaps indicating delivery of postsynaptic F-actin to sites of postsynaptic assembly.
By 24 hours after hatch, we observed clear co-localization of F-actin with newly formed AChR clusters in the dendritic shaft that are positioned for developmental translocation into growing dendritic spines. In mature animals, AChR clusters are stably sequestered at the tips of spines, while F-actin occupies the spine head and neck ( Figure 2H-K).
Dendritic spines form in the absence of presynaptic activity Our above analysis indicated that the localization of synaptic vesicles and active zone proteins, such as CLA-1, occur prior to spine outgrowth. In the rodent brain, spine morphogenesis is clearly regulated by presynaptic activity (Sala & Segal, 2014) (Saneyoshi et al., 2010) (Engert & Bonhoeffer, 1999) (De Roo et al., 2008, but recent evidence suggests initial spine outgrowth may proceed independently of synaptic activity (Lu et al., 2013) (Sigler et al., 2017) (Sando et al., 2017). We therefore next asked whether presynaptic cholinergic activity is important for the formation of DD dendritic spines. To address this question, we analyzed the was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Surprisingly, the abundance of spines and receptor clusters at L4 stage were not significantly affected by disruption of ACh synthesis in cholinergic motor neurons (mutation of the vesicular acetylcholine transporter unc-17) or by a strong reduction in synaptic vesicle exocytosis (mutation of the syntaxin binding protein unc-18). Similarly, loss-of-function mutations in genes required for dense core vesicle release (CAPS/unc-31) and Ca 2+ signaling (VGCC/unc-2, CaM Kinase II/unc-43) did not significantly reduce the number of spines or receptor clusters in the mature circuit (Table 1, Figure S3.1). Our data indicate that alterations in synaptic activity do not affect the abundance of L4 stage dendritic spines. However, we cannot exclude the possibility of developmental delays in spine formation (Cuentas-Condori et al., 2019). Notably, many of the mutations we examined significantly altered spine length and the size of receptor clusters (Table 1), suggesting that activity may influence aspects of spine morphology and post-synaptic structure. Together, our findings point toward a model where initial spine formation and postsynaptic development proceeds without a strong requirement for synaptic activity, while subsequent spine morphogenesis may be affected.

Presynaptic NRX-1 stabilizes postsynaptic components
In previous work, we found that mutations in nrx-1, the sole C. elegans ortholog of neurexin, eliminate DD dendritic spines and disrupt cholinergic receptor clusters in the mature C. elegans circuit (Philbrook et al., 2018), indicating alternate activity-independent mechanisms may be critical for spine formation. Consistent with these findings, we found that mutations of nrx-1 disrupted dendritic calcium responses evoked by presynaptic stimulation of cholinergic motor neurons ( Figure S3.2). To examine possible early involvement of presynaptic NRX-1 in was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. the establishment of spines we asked when NRX-1 first localizes to presynaptic terminals of cholinergic motor neurons. Axonal clusters of NRX-1::GFP are present by 16 hours after hatch ( Figure 3A, B), approximately the same time at which assemblies of the presynaptic scaffold CLA-1 become visible ( Figure 2). Diffuse synaptic vesicle material is also visible at this time ( Figure 2) but has yet to be organized into discrete clusters.
The early arrival of NRX-1 to synaptic terminals could indicate a role in either initial synapse formation or subsequent stabilization and maturation. To distinguish between these possibilities, we analyzed the development of dendritic spines and cholinergic receptor clusters in nrx-1(wy778) null mutants. We quantified spine and AChR cluster number over a similar time course as previously completed for wild type. Surprisingly, we noted that immature spines and receptor clusters were clearly evident in nrx-1 mutants during early development, albeit at slightly reduced density relative to wild type ( Figures 3C-F, S3.3). The density of spines in nrx-1 mutants increased significantly over the next several hours until 24 hours after hatch ( Figure   3C,D). However, after this time, spine density decreased dramatically such that nrx-1 DD dendrites were almost completely devoid of spines by L4 stage, consistent with our previous work (Philbrook et al., 2018). Measurements of spine formation and disassembly in live imaging studies of wild type and nrx-1 mutants offered further support for this conclusion. Mature wild type spines (L4) were remarkably stable over 1-2 hours of recording, but were more dynamic in the developing circuit (16-20 hrs after hatch) ( Figure S3.4, Videos 3-5). Developing spines in nrx-1 mutants were also highly dynamic. Notably, heightened spine dynamics persisted into more advanced developmental stages for nrx-1 mutants compared with wild type animals. For example, at 21-24 hours after hatch, almost 90% of wild type spines were stable over the recording period, while less than 50% of nrx-1 mutant spines remained stable, indicating decreased stability in the absence of NRX-1.
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Within a few hours however, ACR-12 cluster number and intensity decreased significantly in nrx-1 mutant dendrites, and remained low throughout the remainder of development ( Figure   3E-F, S3.3). Our time course analysis indicated that both spines and receptor clusters initially form normally in nrx-1 mutants, but rapidly disappear as animals proceed through development in the absence of NRX-1. Remarkably, we found that the organization of dendritic F-actin was similarly affected by nrx-1 deletion. GFP::UtrCH was arranged similarly in nrx-1 mutants and wild type in early development (16 hrs after hatch), positioned in discrete clusters along the main dendritic process ( Figure 3G). By L4 stage, wild type F-actin was almost entirely associated with dendritic spines. This organization was strikingly altered in L4 stage nrx-1 mutants. GFP::UtrCH was diffusely localized along the length of nrx-1 mutant dendrites, indicating that presynaptic NRX-1 promotes dendritic F-actin assemblies ( Figure 3G). Taken together, our results indicate that presynaptic NRX-1 is dispensable for the earliest stages of postsynaptic assembly and spine formation, but is critical for stabilizing dendritic spines and AChR clusters and promoting their maturation.
Kinesin-3/UNC-104 transports presynaptic neurexin to cholinergic terminals in order to stabilize postsynaptic components To investigate this model further, we sought to determine how presynaptic neurexin is transported to active zones in vivo. Prior work showed that both synaptic vesicles and CLA-1 depend on the Kinesin-3 motor UNC-104 for their delivery to synapses (Hall & Hedgecock, 1991) (Xuan et al., 2017) (Figures S4.1 and 4). In particular, disruption of unc-104 causes an accumulation of synaptic vesicles within neuronal somata and a corresponding loss of synaptic vesicles within axons (Hall & Hedgecock, 1991) (Figure S4.1). Consistent with potential was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ; involvement of UNC-104 in NRX-1 trafficking, we found that UNC-104 and NRX-1 are partially colocalized in cholinergic axons ( Figure S4.2). unc-104(e1265) carries a D1497N mutation in the PH domain of UNC-104 that impairs cargo binding and leads to a severe reduction in UNC-104 abundance in axons (Kumar et al., 2010). NRX-1 endogenously tagged with GFP is strongly visible in neuronal processes within the nerve ring and nerve cords of wild type animals ( Figure S4.3). In contrast, NRX-1::GFP clusters were strikingly decreased in the nerve cords of motor unc-116 did not produce significant accumulation of NRX-1::GFP in cholinergic somata and caused comparatively modest decreases in axonal NRX-1::GFP, demonstrating preferential involvement of Kinesin-3 for NRX-1 transport ( Figure S4.5). Live imaging studies offered additional evidence that synaptic vesicle and NRX-1 trafficking may share a common dependence on UNC-104. We found that NRX-1::GFP trafficking events occurred with similar anterograde and retrograde velocities to SNB-1::GFP (labeling synaptic vesicles) trafficking events, though less frequent ( Figure S4.6, Videos 6 and 7). Prior studies of active zone protein transport have noted these events occur at reduced frequency compared with synaptic vesicle trafficking events (Lipton et al., 2018). Together, our analyses point to a specific requirement for the UNC-104/KIF1A motor in delivery of NRX-1 to presynaptic terminals.
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. The density of spines and AChR clusters were also severely reduced in L4 stage unc-104(e1265) mutants (Figure 4 G-I), similar to nrx-1 mutants ( Figure 4J). These effects were rescued by either native or cholinergic expression of wild type unc-104 in unc-104(e1265) mutants, but not by GABA-or muscle-specific unc-104 expression (Figure 4 G-I), demonstrating that presynaptic UNC-104-mediated transport is critical in cholinergic axons for postsynaptic spine development and receptor localization.
Notably, we observed that several other presynaptic AZ components also require UNC-104 for their delivery. In addition to CLA-1, transport of both ELKS-1/ELKS/CAST and UNC-10/RIM were severely disrupted by mutation of unc-104, indicating that Kinesin-3 cargo selection and transport is key for the synaptic delivery and assembly of several key active zone constituents in these neurons ( Figure 4C-F). In contrast, the synaptic localization of these active zone proteins was not appreciably affected by nrx-1 deletion ( Figure S4.7). We next asked how a failure in delivery of AZ components may contribute to the severe reductions in spines and receptor clusters we observed in unc-104 mutants. We compared postsynaptic organization in strains carrying deletion mutations in either nrx-1, cla-1, unc-10 or elks-1.
Remarkably, only mutation of nrx-1 produced significant decreases in dendritic spines ( Figure   4J), demonstrating that presynaptic ELKS, UNC-10 and CLA-1 are dispensable for postsynaptic development. Our results indicate a specific requirement for NRX-1 in the stabilization and maturation of postsynaptic structures and provide evidence that a failure in synaptic delivery of NRX-1 is a primary causal factor in the postsynaptic structural defects of unc-104 mutants.
The stabilization of mature dendritic spines requires ongoing synaptic delivery of NRX-1 Our results suggest UNC-104 mediated transport positions NRX-1 at the presynaptic terminal in the early stages of synapse formation where it acts to stabilize growing postsynaptic structures, including spines and receptor clusters. We next sought to address whether there is was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. an ongoing requirement for presynaptic UNC-104-dependent transport in the stabilization of mature spines. To address this question, we used the previously characterized temperaturesensitive allele, unc-104(ce782). unc-104(ce782) animals carry a G105E missense mutation in the motor domain of UNC-104 (Edwards et al., 2015). When grown at permissive temperature (13.5°C), unc-104(ce782) axonal synaptic vesicle abundance and animal motility are modestly reduced compared to wild type. In contrast, synaptic vesicles are completely absent from unc-104(ce782) axons following growth at restrictive temperature (20-25°C), and animal motility is severely compromised within 12 hours of a shift to 23°C (Edwards et al., 2015). We raised unc-104(ce782) animals at the permissive temperature (13.5°C) to L4 stage ( Figure 5). We then shifted L4 stage animals to the restrictive temperature (25°C) for 16-20 hrs, and quantified NRX-1::GFP localization and spine density immediately following this shift. Axonal NRX-1::GFP clusters were strikingly decreased in unc-104(ce782) mutants subjected to the temperature shift compared with control animals subjected to the same shift or unc-104(ce782) mutants raised continuously at the permissive temperature ( Figure 5B,C). unc-104(ce782) mutants grown at the permissive temperature had a slightly reduced number of spines overall ( Figure 5E), compared with wild type. A shift to the restrictive temperature at L4 stage also produced a striking reduction in spine density for unc-104(ce782) mutants ( Figure 5D, E), but not for wild type animals subjected to the same temperature shift. We obtained similar results for unc-104(ce782) mutants using an earlier shift to the restrictive temperature (at L3 stage) ( Figure   S5.1). Taken together, our findings indicate an ongoing requirement for UNC-104 transport that extends well beyond the period of initial synapse formation and spine outgrowth. Among the UNC-104 cargoes we investigated, only deletion of nrx-1 produces a significant reduction in spine density at L4 stage. We therefore propose a model where ongoing UNC-104 delivery of presynaptic NRX-1 is critical for postsynaptic maturation and maintenance of mature spines.
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.  elegans motor circuitry. We find that these dendritic spines share many of the hallmark features of vertebrate dendritic spines, but also have some key differences.
(1) C. elegans spines have morphological similarities with mammalian spines and can be classified into comparable morphological categories.
(2) F-actin is localized to spines while tubulin is mainly present in the dendritic shaft. F-actin is a major component of mammalian dendritic spines and is important for spine structural dynamics (reviewed in (Borovac et al., 2018)). (3) Mitochondria and rough was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ; endoplasmic reticulum are positioned near spines, either in the spine neck or at the base of spines in the dendritic shaft. Dendritic organelles contribute to key processes involved in spine formation and spine metabolism (Liu & Shio, 2008). Ribosomes and mitochondria positioned near or within dendritic spines may aid in modifying dendritic spine activity, local protein synthesis, and act as synaptic ATP and Ca 2+ sources (Steward & Levy, 1982 (Hendricks et al., 2012). Therefore, one intriguing possibility is that, similar to the situation in mammals, spines serve to compartmentalize calcium and perhaps other biochemical signals in GABAergic neurons. Spines may also be required in order to achieve the unusual dyadic arrangement of synapses in the ventral nerve cord, where presynaptic specializations of cholinergic motor neurons are positioned for transmission onto both GABA motor neuron and muscle postsynaptic partners (Hall & Russell, 1991). In this case, GABAergic spines may have developed to intercept cholinergic release sites onto muscles, as suggested previously (J. White, 2018).
Ventrally-directed (VA/B) cholinergic neurons are born and integrated into the motor circuit post-embryonically, forming new synaptic connections with both ventral muscles and dorsally-directed (DD) GABAergic motor neurons. We found that F-actin is compartmentalized to discrete regions within the dendritic shafts of DD neurons very early in the development of the circuit (prior to the appearance of spines) and later becomes exclusively localized to dendritic was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ; https://doi.org/10.1101/2021.05.11.443328 doi: bioRxiv preprint spines. F-actin labeling of spines is apparent as soon as they are detectable, suggesting that Factin assemblies may participate in the earliest stages of spine development. Prior to remodeling of DD neurons, sites of GABA release are located on the ventral processes of DD neurons. As F-actin based structures are typically associated with release sites (Chia et al., 2012) (Meng et al., 2015), our finding that F-actin is clustered in ventral DD processes prior to remodeling may suggest that F-actin initially associated with release sites persists in the DD neurite through remodeling and is then repurposed for postsynaptic development, perhaps acting as a landmark or an actin-based mechanism for stimulating post-synaptic maturation.
Notably, we also observed punctate cholinergic receptor fluorescence in GABAergic dendrites several hours prior to the formation of dendritic spines. These immature receptor clusters colocalize with dendritic F-actin assemblies suggesting that receptors are trafficked into dendrites and positioned with F-actin to rapidly populate growing spines. Indeed, we observed that receptor clusters are visible in growing spines as soon as spines can be clearly resolved. This is consistent with recent findings from time-lapse 2-photon imaging of hippocampal organotypic slice cultures showing that receptor accumulation in spines occurs concurrently with spine outgrowth (Zito et al., 2009). Similarly, we found that synaptic material accumulates in presynaptic axons prior to the emergence of spines from dendritic processes. In particular, initial accumulations of the presynaptic scaffold CLA-1 and the synaptic vesicle marker synaptobrevin/SNB-1 were visible in the axon with a similar time-course to the appearance of dendritic cholinergic receptor clusters. Presynaptic material became more clearly localized to discrete puncta over the next 4-6 hours, occurring roughly coincident with receptor accumulation at the tips of growing spines. Notably, our findings do not support a strong requirement for presynaptic neurotransmitter release in spine formation. The abundance of spines was either mildly affected or unaffected across several mutant strains with severely impaired presynaptic function.
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ; https://doi.org/10.1101/2021.05.11.443328 doi: bioRxiv preprint Transsynaptic NRX-1/Neurexin signaling stabilizes dendritic F-actin assemblies to promote postsynaptic maturation Numerous studies across a variety of model systems have examined roles for neurexins in synapse development and function. However, no single consensus view of neurexin function has emerged. Instead, the function of neurexin at specific synapses are thought to be dictated by cellular and molecular context. Mammalian genomes encode three Nrxn genes that can each give rise to a-, b-, and g-Nrxn isoforms. These isoforms share common intracellular and transmembrane regions but differ in their extracellular domains. a-Nrxn has six extracellular LNS domains interleaved with three EGF-like repeats. b-Nrxn has only one LNS domain, while g-Nrxn lacks all identifiable extracellular domains. Conditional triple knockout of mouse neurexin indicated synapse loss in parvalbumin-positive prefrontal cortex interneurons, but not somatostatin-positive interneurons (L. Y. Chen et al., 2017). At mouse calyx synapses, conditional triple neurexin panneuronal knockout indicated a role in active zone organization and Ca 2+ and BK channel regulation, but no requirement in synapse formation or maintenance (Luo et al., 2020). The multi-faceted roles for neurexins at synapses likely emerge as a consequence of the many neurexin isoforms that are generated by alternative splicing in vertebrates, the complexity of their cellular expression, and the potential for these isoforms to selectively interact with distinct postsynaptic partners.
The C. elegans genome encodes a single neurexin gene, nrx-1 which generates two isoforms, a long a and a short g isoform (Haklai-Topper et al., 2011). As is the case in other systems, C. elegans NRX-1 has roles in presynaptic organization, for example in calcium channel clustering at the active zone (Kurshan et al., 2018). Interestingly, the short g-NRX-1 isoform that lacks identifiable ectodomains performs these roles (worms do not encode b-NRX-was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ; 1), but even in the absence of both isoforms, a level of presynaptic functionality is maintained, as indicated by the presence of evoked responses in muscle cells following presynaptic motor neuron stimulation (Philbrook et al., 2018). Our work here and our prior studies pointed toward the importance of NRX-1 at synapses between cholinergic and GABAergic motor neurons (Philbrook et al., 2018). Surprisingly, our time course studies here revealed that initial spine outgrowth and AChR clustering occurs normally in nrx-1 mutants. However, spines and AChRs in nrx-1 mutants become destabilized within hours and collapse, such that GABAergic dendrites are almost completely devoid of these postsynaptic components by L4 stage. Dendritic F-actin assemblies, that normally show discrete localization to spines, are also disorganized in nrx-1 mutants, and are diffusely distributed throughout GABAergic dendritic processes in nrx-1 mutants. We propose a model where presynaptic NRX-1 is required for the stabilization of dendritic spines and other postsynaptic structures in GABAergic dendrites. We suggest that presynaptic NRX-1 directs the organization of dendritic F-actin to stimulate maturation and stabilization of spines and postsynaptic receptor clusters. A similar form of F-actin based reorganization is a key step in presynaptic differentiation (Chia et al., 2012) (Chia et al., 2014).
Our work has parallels with recent studies in Xenopus (S. X. Chen et al., 2010) and Drosophila (Constance et al., 2018). In embryonic Xenopus brain, presynaptic b-neurexin stabilizes dendritic filopodia through an adhesive partnership with neuroligin to direct the development of dendritic arbors (S. X. Chen et al., 2010). Similarly, during fly metamorphosis, neurexin/neuroligin-based adhesion promotes the growth of neurite branches independently of synaptic activity (Constance et al., 2018). In our studies, NRX-1 stabilization of dendritic spines also occurs independently of synaptic activity and without involvement of neuroligin (Philbrook et al., 2018), suggesting interaction with an alternate binding partner that to date remains unidentified.
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ; https://doi.org/10.1101/2021.05.11.443328 doi: bioRxiv preprint UNC-104/Kinesin 3-mediated synaptic delivery of NRX-1/Neurexin is an essential early step required for synapse maturation and stabilization Efficient trafficking of synaptic and active zone cargoes to presynaptic terminals is essential for synapse formation and neuronal communication. Long-range axonal transport of synaptic vesicles and active zone proteins is carried out by anterograde kinesin and retrograde dynein motors. However, significant questions remain about which synaptic proteins are trafficked together and their dependence on specific motors. We found a specific requirement for the kinesin-3 UNC-104/KIF1A motor to deliver NRX-1 to presynaptic terminals. The anterograde transport of synaptic and dense core vesicles are also strongly dependent upon UNC-104 (Hall & Hedgecock, 1991) (Zahn et al., 2004. In contrast, several studies suggest that other AZ proteins exhibit weaker requirements for UNC-104 in their delivery. For example, ELKS-1 and other AZ proteins localize to C. elegans HSN synapses in an UNC-104independent manner (Patel et al., 2006). Similarly, UNC-10/RIM properly concentrates within the nerve cord of unc-104(e1265) mutants, while synaptic vesicles accumulate in neuronal cell bodies (Koushika et al., 2001). Here, we found that NRX-1, UNC-10/RIM, ELKS-1/ELKS and CLA-1/piccolo all require UNC-104 for anterograde transport to cholinergic motor neuron synapses suggesting cell-specific functions for UNC-104 mediated delivery of active zone molecules.
Prior studies of rodent cultured neurons also suggested a requirement for the Kinesin-3 motor in neurexin transport, indicating that mechanisms for NRX-1 delivery to synapses are conserved (Neupert et al., 2015). The effects of a failure in delivery on synapse structure and function were not previously assessed. We found that dendritic spines collapse and AChR clusters disperse when UNC-104 delivery of synaptic cargoes is disrupted. Of the UNC-104 cargoes we analyzed, only mutation of nrx-1 produced significant disruption of dendritic spines.
We therefore propose that NRX-1 is a key UNC-104 cargo required for spine stabilization. Our was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ; https://doi.org/10.1101/2021.05.11.443328 doi: bioRxiv preprint live imaging points to similar rates of transport for synaptic vesicles and NRX-1. However, it remains unclear whether NRX-1 is delivered as a component of synaptic vesicles or may be segregated into a distinct vesicular population, such as synaptic vesicle protein transport vesicles (STVs). Our findings that mutations which severely impair synaptic vesicle release do not affect spine density point toward the latter possibility.
The prior identification of a temperature-sensitive allele of unc-104 allowed us to explore the temporal requirements for UNC-104 delivery of NRX-1. Importantly, we found that UNC-104 delivery was not solely required during early synaptogenesis but was critical throughout the developmental progression of the circuit toward maturity. These findings argue that ongoing delivery of presynaptic NRX-1 is required to maintain postsynaptic structure long after circuit assembly is complete. This raises interesting questions about the relationship between NRX-1 transport and synapse stability, perhaps suggesting that alterations in the rate of axonal NRX-1 transport may directly impact synaptic connectivity in mature animals. More broadly, our studies of the development and stabilization of synapses at dendritic spines in C. elegans provide a new view of the role of adhesive mechanisms in circuit connectivity and highlight a novel role for neurexin in the stabilization of mature synapses and dendritic spines.
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Strains
All strains are N2 Bristol strain derivatives (wild type) and were maintained at room temperature (20-24°C) on nematode growth media plates (NGM) seeded with E. coli strain OP50.
Transgenic strains were obtained by microinjection to achieve transformation (Mello et al., 1991) and identified using co-injection markers. Integrated lines were produced with X-ray irradiation and outcrossed to wild type/N2 Bristol eight times. Only hermaphrodites (L1-L4, young adults) were used in this study. A complete list of all strains used in this study is found in Supplemental File 1. Worms used for time course studies were staged by transferring embryos to seeded OP50 plates and transferring to 25°C (time point 0).

Molecular Biology
Plasmids were constructed using the two-slot Gateway Cloning system (Invitrogen) and confirmed by restriction digest and sequencing.
TBA-1/tubulin reporter. Sequence coding for GFP::TBA-1 was amplified from plasmid pYJ128 (gift from Kang Shen laboratory) and ligated into a destination vector to create pDest-173.
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Confocal microscopy
All strains were immobilized with sodium azide (0.3 M) on a 2% or 5% agarose pad. Images were obtained using an Olympus BX51WI spinning disk confocal equipped with a 63x objective.
For time course analyses, newly hatched larvae were transferred to a seeded OP50 plate and maintained at 25°C.
Long term live imaging of dendritic spines and F-actin. Nematodes were immobilized using 2 µL 50 mM muscimol in 10 µL 20% PEG hydrogel solution. Once paralyzed, hydrogel was curated using a handheld UV Transilluminator (312 nm, 3 minutes). Z-stacks were acquired every 5 minutes for at least one hour.
Live imaging of synaptic vesicle and NRX-1 trafficking. Nematodes were immobilized using 50 mM muscimol on a 5% agarose pad. Cholinergic commissures were imaged using Perkin Elmer spinning disk confocal equipped with a 63x objective using 100 ms exposure for 30 seconds.

Confocal microscopy analysis
All image analysis was conducted using ImageJ software (open source) within defined ROIs using intensity threshold values determined from control experiments. ROIs were located 15-30 µm anterior to DD1, DD2 or DD3 somata Spine analysis. Mature spines were quantified as protrusions from the dendrite >0.3 µm in length (measuring from the base to the tip of the protrusion). Spine density was defined as the number of spines/unit length within a selected ROI.
3D rendering was conducted using Imaris/bitplane 3D image analysis software. Morphological categories were determined based on criteria used in (Harris et al., 1992). The ImageJ line tool was used to measure the length and width of the spine 3D rendering. Based on these calculations, spines were placed into one of four morphological categories based on the was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.  For soma fluorescence intensity, ROI was drawn around the perimeter of the soma.
Synaptic marker and receptor cluster. Background fluorescence was subtracted, and the number and size of synaptic/receptor puncta were measured using the 'analyze particles' function. Confocal montages were assembled using the 'straighten to line' function in a 50 µm region of the dorsal nerve cord.

Calcium Imaging
Animals were grown on plates seeded with OP50 containing 2.7 mM All-Trans Retinal (ATR).
Plates were stored at 4°C under dark conditions and used within one week. Imaging was carried out using L4 animals immobilized in hydrogel (Burnett et al., 2018). Animals were transferred to 7.5 µL of the hydrogel mix placed on a silanized glass slide and covered with a glass slide.
Hydrogel was cured using a handheld UV Transilluminator (312 nm, 3 minutes). Post-curing, the covering slide was removed and replaced with a coverslip. Imaging was carried out using a Yokogawa CSU-X1-A1N spinning disk confocal system (Perkin Elmer) equipped with EM-CCD camera (Hamamatsu, C9100-50) and 63X oil immersion objective. Chrimson photoactivation (~30 mW/cm 2 ) was achieved using a TTL-controlled 625 nm light guide coupled LED (Mightex Systems). A 556 nm BrightLine single-edge short-pass dichroic beam splitter was positioned in the light path (Semrock) (Figure S1.3). Data were acquired at 10 Hz for 15 s using Volocity software and binned at 1×1 during acquisition. Analysis was performed using ImageJ. The DD was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ; https://doi.org/10.1101/2021.05.11.443328 doi: bioRxiv preprint neurite process in each time series was extracted using the straighten function, background subtracted and photobleaching correction was carried out by fitting an exponential function to the data (CorrectBleach plugin). A smoothing function was applied to the data to enhance signal-to-noise. Individual spine ROIs were identified using the mCherry fluorescence. Post imaging processing, pre-stimulus baseline fluorescence (F0) was calculated as the average of the data points in the first 4 s of the recording. Data was normalized to prestimulus baseline as ∆F/F0, where ∆F = F -F0. Peak ∆F/F0 was determined by fitting a Gaussian function to the ∆F/F0 time sequence using Multi peak 2.0 (Igor Pro, WaveMetrics). All collected data were analyzed, including failures (no response to stimulation). Control recordings were carried out in the absence of Retinal.
We identified and characterized 6 DD1 spines from a 35 µm region of the ventral head region where the neurite emerges from the cell body of DD1 GABAergic motor neuron.

Temperature shift experiments
Wild type and unc-104(ce782)ts animals eggs were raised at 13.5°C until L3 or L4 stage as indicated, then shifted to 25°C for 16-20 hrs prior to imaging. Animals were age-matched at the time of the temperature shift to account for developmental delays of unc-104(ce782)ts mutants.
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ; https://doi.org/10.1101/2021.05.11.443328 doi: bioRxiv preprint resources of polyribosomes and SER promote synapse enlargement and spine clustering after long-term potentiation in adult rat hippocampus. Sci Rep,9(1) Synapse-Assembly Proteins Maintain Synaptic Vesicle Cluster Stability and was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.  was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.   was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.  Numbering corresponds to representative traces in Figure 1I.
(I) Representative evoked responses for indicated dendritic spines in wild type animals (wild type 1, 2 and 3, Figure 1H), animals grown in the absence of retinal was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ; https://doi.org/10.1101/2021.05.11.443328 doi: bioRxiv preprint was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ; https://doi.org/10.1101/2021.05.11.443328 doi: bioRxiv preprint was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ;   Table 1 Spines and AChR localize to GABAergic neurons independent of presynaptic activity. was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ; was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021  was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ; Images on each line are from different animals (5 shown for each genotype).
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

(A)
Cartoon depiction of experimental timeline. Animals were grown at 13.5°C until L4 stage (approx. 144 hours in unc-104(ce782)ts mutants, 120 hours in wild type was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ; https://doi.org/10.1101/2021.05.11.443328 doi: bioRxiv preprint animals) before shifting animals to the restrictive temperature of 25°C for 16-20 hours and imaging. was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.   Figure S1.4B. Scale bar, 100 nm.
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ; https://doi.org/10.1101/2021.05.11.443328 doi: bioRxiv preprint   as not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ; https://doi.org/10.1101/2021.05.11.443328 doi: bioRxiv preprint Figure S1.3. Cartoon illustration of calcium imaging recording platform. Imaging was carried out using a Yokogawa CSU-X1-A1N spinning disk confocal system (Perkin Elmer) equipped with EM-CCD camera (Hamamatsu, C9100-50) and 63X oil immersion objective. Chrimson photoactivation (~30 mW/cm 2 ) was achieved using a TTLcontrolled 625 nm light guide coupled LED (Mightex Systems), permitting illumination of the entire immobilized animal, while simultaneously recording GCaMP6f fluorescence (excitation 488 nm, emission 525 nm). 625 nm light was prevented from reaching the was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ; https://doi.org/10.1101/2021.05.11.443328 doi: bioRxiv preprint was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021  was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021  Hours/Age post-hatch spine length (μm) C was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021 to DD1 soma. Blue dashed rectangle indicates inset anterior to DD2 soma. Note that DD1 spines form prior to DD2 spines. (C) Quantification of the length of growing dendritic spines at 24 and 32 hours after hatch and at L4 (~42-50 hours after hatch) stage. Data points indicate mean ± SEM.
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021  was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021  was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021 Video 1: Confocal live imaging video of F-actin dynamics (labeled by  in the DD1 dendrite of wild type at 16 hours after hatch. For display, movies are shown at 3 fps. Images were acquired every 5 minutes.
Video 2: Confocal live imaging video of F-actin dynamics (labeled by  in the DD1 dendrite of wild type at L4 stage. For display, movies are shown at 3 fps. Images were acquired every 5 minutes. was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021  was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Bars indicate mean ± SEM. One-way ANOVA, Dunnett's multiple comparisons, ****p<0.0001. Wild type control is the same as Figure 1J. n ≥ 10 animals. was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021   was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.  was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.  Distance (μm) Relative intensitŷ^^^^^A B was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Note that NRX-1 trafficking events occur significantly less frequently than SNB-1 events.
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted May 11, 2021. ; https://doi.org/10.1101/2021.05.11.443328 doi: bioRxiv preprint was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.