Plasticity-induced actin polymerization in the dendritic shaft regulates intracellular AMPA receptor trafficking

Abstract
Editor's evaluation
Introduction
Results
Discussion
Materials and methods
Appendix 1
Data availability
References
Article and author information
Metrics

Abstract

AMPA-type receptors (AMPARs) are rapidly inserted into synapses undergoing plasticity to increase synaptic transmission, but it is not fully understood if and how AMPAR-containing vesicles are selectively trafficked to these synapses. Here, we developed a strategy to label AMPAR GluA1 subunits expressed from their endogenous loci in cultured rat hippocampal neurons and characterized the motion of GluA1-containing vesicles using single-particle tracking and mathematical modeling. We find that GluA1-containing vesicles are confined and concentrated near sites of stimulation-induced structural plasticity. We show that confinement is mediated by actin polymerization, which hinders the active transport of GluA1-containing vesicles along the length of the dendritic shaft by modulating the rheological properties of the cytoplasm. Actin polymerization also facilitates myosin-mediated transport of GluA1-containing vesicles to exocytic sites. We conclude that neurons utilize F-actin to increase vesicular GluA1 reservoirs and promote exocytosis proximal to the sites of synaptic activity.

Editor's evaluation

In this manuscript, the authors developed a sensitive single particle tracking method for endogenous AMPA receptors. They found that AMPAR-containing vesicles showed reduced mobility near stimulation sites, due to increased F-actin bundling in dendritic shafts. The study provides compelling evidence on a new important mechanism of AMPAR trafficking using state-of-the-art labeling and analysis techniques.

https://doi.org/10.7554/eLife.80622.sa0

Introduction

Synaptic plasticity – the modulation of synaptic connections in response to changes in neuronal activity – is regarded as the cellular basis for learning and memory (Abbott and Nelson, 2000; Citri and Malenka, 2008; Magee and Grienberger, 2020). Changes in postsynaptic spine morphology (i.e. structural plasticity) and protein composition are observed during synaptic plasticity and are thought to contribute to changes in synaptic transmission (i.e. functional plasticity; Kasai et al., 2010; Nakahata and Yasuda, 2018; Kasai, 2023). Persistent pre-synaptic neurotransmitter release stimulates postsynaptic spine enlargement and insertion of proteins into the synapse (Matsuzaki et al., 2004; Okamoto et al., 2004; Hayashi and Majewska, 2005; Patterson and Yasuda, 2011), including α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPARs; Song and Huganir, 2002) – ionotropic glutamate receptors that permit the influx of sodium, potassium, and in certain cases calcium ions (Chater and Goda, 2014). This increase in AMPAR abundance, as well as changes to AMPAR conductance, results in a greater inward flow of ions in response to stimulation (Anggono and Huganir, 2012; Diering and Huganir, 2018).

Given that the number of AMPARs at synapses is highly regulated during synaptic plasticity, AMPAR trafficking has been studied extensively (Malinow and Malenka, 2002; Henley et al., 2011; Park, 2018; Diering and Huganir, 2018; Groc and Choquet, 2020; Díaz-Alonso and Nicoll, 2021). The majority of AMPARs are synthesized in the cell body and then enter the secretory pathway, where they are trafficked in vesicles and inserted into the neuronal membrane via exocytosis (Shepherd and Huganir, 2007). AMPARs on the membrane can diffuse into synapses (Choquet and Opazo, 2022) and can also be endocytosed and delivered to lysosomes for degradation or recycled through the endocytic pathway (Ehlers, 2000; Carroll et al., 2001; Goo et al., 2017; Hanley, 2018). An important feature of synaptic plasticity is input specificity – plasticity stimulated at one synapse does not spread to inactive synapses (Kandel et al., 2013) – suggesting that AMPARs are inserted selectively into stimulated synapses. The mechanisms that regulate how AMPARs are delivered to synapses undergoing plasticity are not fully understood, but evidence points to two models: (1) AMPARs in vesicles are trafficked and exocytosed directly into or adjacent to stimulated synapses (Gerges et al., 2006; Kennedy et al., 2010; Cho et al., 2015; Park, 2018); and (2) AMPARs diffusing in the neuronal membrane are selectively trapped at stimulated synapses (Borgdorff and Choquet, 2002; Tardin et al., 2003; Petrini et al., 2009; Ehlers et al., 2007; Opazo et al., 2010; Opazo and Choquet, 2011; Penn et al., 2017; Choquet and Opazo, 2022).

AMPARs in vesicles are trafficked through dendrites by both microtubule- and actin-based motors, and disrupting active transport of AMPARs interferes with long-term potentiation (LTP), a commonly studied form of plasticity (Setou et al., 2002; Hoogenraad et al., 2005; Correia et al., 2008; Wang et al., 2008; Hoerndli et al., 2013; Hoerndli et al., 2015; Wagner et al., 2019; Gutiérrez et al., 2021). Nevertheless, it is unclear whether AMPAR-containing vesicles (hereafter referred to as AMPAR vesicles) are delivered directly to synapses undergoing plasticity, due in large part to limitations in imaging AMPARs in vesicles as opposed to on cell membranes (Groc and Choquet, 2020). One approach to studying AMPAR vesicles utilizes chemically inducible dimerization to control the release of exogenous AMPARs (i.e. AMPARs expressed from plasmid DNA) from the endoplasmic reticulum into the secretory pathway, followed by tracking AMPAR vesicles as they traverse photobleached sections of dendrite (Hangen et al., 2018; Bonnet et al., 2023). Using this technique, Hangen et al., 2018 found AMPAR vesicles slow down and pause in response to elevated intracellular calcium levels during neuronal activity, and consequently hypothesized that a calcium-mediated mechanism primes AMPAR vesicles for exocytosis. However, it is unclear if pausing is directly linked to AMPAR exocytosis at synapses undergoing plasticity. Additional studies have demonstrated that endosomes containing exogenous AMPARs can enter dendritic spines (EstevesdaSilva et al., 2015; Bowen et al., 2017), raising the possibility of direct AMPAR exocytosis into synapses. However, scanning electron micrographs of immunogold-labeled AMPARs fail to reveal a substantial fraction of AMPAR vesicles in spines (Tao-Cheng et al., 2011). Moreover, imaging exogenous AMPARs tagged with super ecliptic pHluorin shows that exocytosis occurs largely at extrasynaptic sites (Lin et al., 2009; Makino and Malinow, 2009; Patterson et al., 2010), favoring a model in which receptors diffuse into synapses after exocytosis.

Much research has been focused on understanding how synapses capture AMPARs as they diffuse through the neuronal membrane (Opazo and Choquet, 2011; Groc and Choquet, 2020; Choquet and Opazo, 2022). After exocytosis, AMPARs diffuse freely in random directions, but diffusion decreases precipitously at synapses (Tardin et al., 2003) because AMPARs are anchored there by postsynaptic density (PSD) proteins, such as PSD-95 (El-Husseini et al., 2000; Bats et al., 2007; Opazo et al., 2010; Opazo et al., 2012; Chen et al., 2015). Synaptic activity changes both the composition of proteins in the synapse and posttranslational modifications on AMPARs, further enhancing receptor anchoring (Opazo et al., 2010; Diering and Huganir, 2018; Lu and Roche, 2012; Opazo et al., 2012). These observations support a model where AMPARs may not be trafficked to specific loci, but rather diffuse in random directions, only to be concentrated in active synapses as a consequence of their increased residence time. Importantly, crosslinking AMPARs on the neuronal membrane to prevent their diffusion impairs synaptic potentiation in vivo (Penn et al., 2017; Getz et al., 2022). Nevertheless, the net distance a receptor can travel via diffusion is limited (Groc and Choquet, 2020). Consequently, this model depends on the presence of nearby extrasynaptic reservoirs from which synapses can draw AMPARs during plasticity (Choquet and Opazo, 2022). How reservoirs are established and maintained is not fully understood, but given that synapses can be located hundreds of microns from the cell body, it is probable that receptors are actively transported near sites undergoing synaptic plasticity.

To address whether and how neurons specify the location to which AMPAR vesicles are delivered, we developed a method to identify vesicles containing AMPAR GluA1 subunits expressed at native levels from endogenous loci and characterize the motion of these vesicles in cultured rat hippocampal neurons. Using this technique, we identify previously undescribed motion behaviors for GluA1-containing vesicles (hereafter referred to as GluA1 vesicles). We show that stimulating synaptic activity with glycine-induced chemical LTP (cLTP) or structural plasticity with glutamate uncaging-evoked structural LTP (sLTP) results in the local confinement of GluA1 vesicles in the dendritic shaft. We find that confinement concentrates GluA1 vesicles near sites of stimulation, thereby increasing the size of GluA1 reservoirs near these sites. GluA1 vesicle confinement is the result of stimulation-induced actin polymerization in the dendritic shaft, which changes the rheological properties of the dendritic cytoplasm in a manner that inhibits transport along the length of the dendrite and inhibits diffusion of GluA1 vesicles. Finally, we show that actin polymerization in the dendritic shaft near the sites of stimulation facilitates myosin-mediated transport of GluA1 vesicles from intracellular reservoirs to sites of exocytosis. In sum, our results suggest that neurons enhance the delivery AMPARs to active synapses by restricting the motion of AMPAR vesicles away from these synapses while simultaneously promoting AMPAR exocytosis near these synapses.

Results

GluA1-HaloTag is trafficked to postsynaptic densities and responds to stimulation

To study the intracellular transport of AMPARs during neuronal activity, we developed a method to label endogenous AMPAR GluA1 subunits (encoded by Gria1), expressed at native levels, using homology-independent targeted integration (HITI; Suzuki et al., 2016) in cultured rat hippocampal neurons. HaloTag (HT; Los et al., 2008) enables labeling of GluA1 with bright and photostable Janelia Fluor (JF) dyes (Grimm et al., 2015), which can be conjugated to the HaloTag ligand (HTL). We used HITI to insert HaloTag into the extracellular amino-terminal domain (NTD) of GluA1 at R280 (Figure 1A and Figure 1—figure supplement 1A), resulting in a high knock-in efficiency (Figure 1—figure supplement 1B–C). GluA1 edited at this position with HaloTag (GluA1-HT) and labeled with JF₅₄₉-HaloTag ligand (JF₅₄₉-HTL) is concentrated in dendritic spines, similar to endogenous GluA1 (Figure 1B; Craig et al., 1993). We find that the majority of edited sequences contain HaloTag in the correct orientation and without indels (Figure 1—figure supplement 1D), and that HaloTag insertion does not alter the expression of Gria1 (Figure 1—figure supplement 2A–B). We observe correlated localization of HaloTag and Gria1 mRNA in the same neuron (Figure 1—figure supplement 2C), indicating that Gria1-HaloTag is produced as a single transcript. Using stimulation emission depletion (STED) microscopy to examine GluA1 and HaloTag labeling at subdiffraction-limited length scales, we find that HaloTag signal has a strong overlap with native GluA1 signal (Figure 1—figure supplement 3A), indicating that GluA1-HT is translated as a single peptide sequence. Using immunofluorescence labeling and STED, we demonstrate that HaloTag insertion at R280 does not disrupt GluA1 trafficking to postsynaptic densities (Figure 1—figure supplement 3B–D). Finally, using electrophysiological methods, we demonstrate that insertion of HaloTag at R280 in the NTD of GluA1 does not significantly alter its channel function (Figure 1—figure supplement 4).

Figure 1 with 17 supplements see all

Download asset Open asset

Chemical LTP induction reduces active transport and diffusion of GluA1-HT vesicles in the dendritic shaft.

(A) Schematic of *Gria1* gene targeting with HaloTag (HT) by homology-independent targeted integration (HITI). A donor construct for HaloTag integration (Donor) is co-transfected into rat hippocampal neurons with a construct expressing Cas9 (Cas9). The donor contains HaloTag flanked on each side by one copy of the *Gria1* sequence to be targeted by Cas9, a single guide RNA (sgRNA), and a neuronal transfection marker (e.g. miRFP670). In neurons transfected with both the Cas9 and Donor constructs, Cas9 creates a double strand break in the genomic copy of *Gria1* and also excises the HaloTag sequence from the donor construct. The freed HaloTag sequence can be inserted into the genomic cut site when the double strand break is repaired by non-homologous end joining (NHEJ). (B) Representative confocal image of a cultured rat hippocampal neuron expressing endogenous GluA1 tagged with HaloTag and labeled with JF₅₄₉-HaloTag ligand (GluA1-HT-JF₅₄₉). Scale bar, 20 μm. (C) Experimental workflow to achieve sparse GluA1-HT labeling for GluA1-HT vesicle identification and single-particle tracking (SPT) analysis. (D) GluA1-HT vesicle trajectories in a dendritic shaft with no treatment (Control) and during cLTP induction (cLTP). Trajectories are overlaid on epifluorescence images of GFP-Homer1c, which was used to identify and segment the dendritic shaft. Scale bar, 10 μm. For videos, see Figure 1—video 5 and Figure 1—video 6. (E) Pie charts: fractions of vesicles exhibiting diffusion or active transport in dendritic shafts with no treatment (Control) and during cLTP induction (cLTP). ****p<0.0001 by Mann-Whitney test. n=12–14 timelapse imaging sequences (each timelapse captures the motion of GluA1-HT vesicles in one region of dendrite in one neuronal culture) for each condition. Histogram: distributions of diffusion coefficients for GluA1-HT vesicles without treatment (Control; gray) and during cLTP induction (cLTP; blue). Line represents the probability density function of each histogram estimated by kernel density estimation (KDE). ****p<0.0001 by Kolmogorov-Smirnov test. n=227–360 GluA1-HT vesicle trajectories pooled from 12 to 14 timelapses for each condition.

Figure 1—source data 1 Related to Figure 1.: https://cdn.elifesciences.org/articles/80622/elife-80622-fig1-data1-v1.xlsx
Download elife-80622-fig1-data1-v1.xlsx

Neuronal activity confines GluA1 vesicles by disrupting vesicle motion in the dendritic shaft of cultured rat hippocampal neurons

To overcome the challenge of signal saturation (Figure 1B) and track single GluA1-HT vesicles inside the dendritic shaft, we developed a block-and-chase protocol to achieve sparse labeling of de novo synthesized endogenous GluA1 (Figure 1C). To avoid visualizing pre-existing GluA1-HT, GluA1-HT was first labeled with a saturating concentration of JF₆₄₆-HTL in the presence of the translation inhibitor cycloheximide (CHX). Next, JF₆₄₆-HTL and CHX were washed away, and after an incubation period to allow GluA1-HT translation to recover, de novo synthesized GluA1-HT was labeled with JF₅₄₉-HTL. Because JF₅₄₉-HTL labels only newly synthesized GluA1-HT, signal detected by fluorescence microscopy is dramatically reduced in this fluorescence channel, allowing us to identify sparse, punctate GluA1-HT-JF₅₄₉ conjugates (Figure 1—video 1). We then used single-particle tracking (SPT) analysis to reconstruct trajectories, and hidden Markov modeling with Bayesian model selection (HMM-Bayes; Monnier et al., 2015) to determine important motion parameters (i.e. velocity and diffusion coefficient) and predict the motion type (i.e. active transport versus diffusion) of each trajectory (Figure 1—figure supplement 5; Jaqaman et al., 2008; Liu et al., 2018). Finally, we separated GluA1-HT vesicles from surface GluA1-HT based on motion type, fluorescence bleaching characteristics, and localization in the dendritic shaft (Figure 1—figure supplements 6–8).

Having established a pipeline to label, track, and analyze GluA1-HT vesicles, we sought to evaluate whether the motion of GluA1-HT vesicles is modulated by synaptic activity. Glycine-induced chemical LTP (cLTP) is an established method of stimulating synaptic plasticity and increasing the surface expression of GluA1 in cultured neurons (Lu et al., 2001; Passafaro et al., 2001; Park et al., 2006; Molnár, 2011). We hypothesized that cLTP induction might change the motion of GluA1-HT vesicles in the dendritic shaft to support increased GluA1-HT exocytosis during neuronal activity. Using a membrane impermeable variant of JF₅₄₉-HTL termed JF₅₄₉i-HaloTag ligand (JF₅₄₉i-HTL; ‘i’ for impermeant; Xie et al., 2017), we validated that cLTP stimulates increased expression of GluA1-HT on neuronal surfaces (Figure 1—figure supplement 9). We then imaged and tracked GluA1-HT vesicles in dendrites during cLTP induction, and observed clear qualitative differences in motion compared to vesicles in the dendrites of unstimulated control neurons (Figure 1D, Figure 1—video 5 and Figure 1—video 6). Most strikingly, we observed a loss in long-range motion along the length of the dendritic shaft in cLTP-stimulated neurons. When HMM-Bayes is applied to characterize trajectories collected during cLTP induction, we find a significant decrease in the fraction of GluA1-HT vesicles undergoing active transport (Figure 1E, pie charts). The diffusion coefficients (D) for GluA1-HT vesicles undergoing diffusion are also significantly reduced by cLTP induction (Figure 1E, histogram). In addition, vesicles exhibit increased subdiffusion (i.e. constrained diffusion due to molecular crowding or interactions; Feder et al., 1996; Saxton, 2007) in response to cLTP induction (Figure 1—figure supplement 10).

These observations demonstrate that the overall motion of GluA1-HT vesicles is inhibited by cLTP, suggesting that vesicle motion is locally confined (defined in this work as the restriction of vesicle motion away from its initial position). We reasoned that if cLTP spatially confines GluA1-HT vesicles then it should, in addition to decreasing the fraction of vesicles undergoing active transport, prevent diffusing vesicles from transitioning to active transport and leaving their local regions. A critical feature of HMM-Bayes is the ability to infer multiple motion states from a single trajectory and determine state-transition probabilities (Figure 1—figure supplement 11A–B; Monnier et al., 2015). Using HMM-Bayes, we find multi-state GluA1-HT vesicles stochastically switch motion states from active transport to diffusion and from diffusion to active transport with approximately the same probability under unstimulated control conditions (Figure 1—figure supplement 11C, Control, k_AT-D vs k_D-AT). By contrast, during cLTP, GluA1-HT vesicles have a greater probability of switching from active transport to diffusion than from diffusion to active transport (Figure 1—figure supplement 11C, cLTP, k_AT-D vs k_D-AT). Furthermore, GluA1-HT vesicles undergoing diffusion have a high probability to continue diffusing (Figure 1—figure supplement 11C, cLTP, k_D-D). Taken together, our observations suggest that cLTP induction results in the local confinement of GluA1-HT vesicles.

Local induction of synaptic activity confines GluA1 vesicles near the site of activity by disrupting GluA1 vesicle motion

If confinement is a mechanism to increase the intracellular reservoir of GluA1-HT near sites of synaptic activity, then stimulating plasticity at a specific synapse should alter GluA1-HT vesicle motion near that synapse. Single-photon (1 P) 4-Methoxy-7-nitroindolinyl-caged-ʟ-glutamate (MNI) uncaging can be used to stimulate synaptic activity in a desired region (Ellis-Davies, 2007). We used a glutamate-uncaging evoked structural LTP protocol (sLTP; Matsuzaki et al., 2004) to induce structural plasticity at a spine of interest. This protocol has been shown to stimulate N-methyl-ᴅ-aspartate receptor (NMDAR)-mediated calcium transients, which result in spine expansion and increased AMPAR concentration at the targeted synapse – two important proxies for functional plasticity (Matsuzaki et al., 2001; Matsuzaki et al., 2004; Lee et al., 2009; Patterson and Yasuda, 2011; Huganir and Nicoll, 2013; Bosch et al., 2014; Kruijssen and Wierenga, 2019). First, we calibrated the strength of the laser so that it would not trigger calcium influx or structural plasticity in neighboring spines. In targeted spines, we observed a significant increase in the area of the spine head and in GluA1-HT after stimulation in dishes containing MNI, but not in dishes without MNI (Figure 2A and Figure 2—figure supplements 1–2). The increase in spine size persists 10 min after the cessation of sLTP, indicating that this protocol induces sustained changes to activity (Figure 2—figure supplement 3A, images and Spine area line graph).

Figure 2 with 5 supplements see all

Download asset Open asset

Structural LTP stimulation reduces active transport and diffusion of GluA1-HT vesicles proximal to the site of synaptic activity.

(A) Representative epifluorescence images of a dendritic spine expressing GFP immediately before sLTP (Pre-sLTP) and after sLTP (Post-sLTP). Yellow dot indicates the site of uncaging. Scale bar, 2 μm. (B) GluA1-HT vesicle trajectories immediately before sLTP (Pre-sLTP) and after sLTP (Post-sLTP) overlaid on a dendrite expressing GFP. Dendrite is separated into three equal zones based on proximity to the stimulated spine to distinguish proximal and distal areas (Materials and methods). Insets: magnified images of trajectories near the site of uncaging (yellow dot). Scale bars, 5 μm. For video, see Figure 2—video 1. (C) Fractions of GluA1-HT vesicles exhibiting active transport pre- and post-sLTP in each zone. **p=0.0039 by Wilcoxon matched pairs test. Each dot represents the fraction of vesicles exhibiting active transport in the indicated zone for one sLTP stimulation experiment. n=9 sLTP stimulation experiments (each experiment targets one spine on one neuron) where GluA1-HT vesicle motion in the dendrite is captured immediately before and after sLTP stimulation. (D) Distributions of diffusion coefficients for GluA1-HT vesicles in Zone 1 (sLTP 1; blue) versus Zone 2+3 (sLTP 2+3; gray) after sLTP stimulation. Line represents the probability density function of each histogram estimated by kernel density estimation (KDE). **p=0.0037 by Kolmogorov-Smirnov test. n=60–66 GluA1-HT vesicle trajectories pooled together from nine sLTP stimulation experiments.

Figure 2—source data 1 Related to Figure 2.: https://cdn.elifesciences.org/articles/80622/elife-80622-fig2-data1-v1.xlsx
Download elife-80622-fig2-data1-v1.xlsx

To examine GluA1-HT vesicle motion proximal to the site of structural plasticity, we separated the dendritic shaft longitudinally (i.e. along the length of the dendrite) into three equal zones based on proximity to the site of uncaging and assessed the different types of motion that GluA1-HT vesicles exhibit in each zone after sLTP (Figure 2B and Figure 2—video 1). We find that sLTP results in reduced active transport and lower rates of diffusion for GluA1-HT vesicles in Zone 1 but not Zone 2 or Zone 3 (Figure 2C–D), indicating that stimulation disrupts GluA1-HT vesicle motion proximal, but not distal, to the site of synaptic activity. By contrast, sLTP stimulation in the absence of MNI failed to alter the fraction of vesicles exhibiting active transport or the diffusion coefficient of vesicles both proximal and distal to the site of stimulation (Figure 2—figure supplement 3B–C), demonstrating that changes in vesicle movement after sLTP are not the result of laser exposure. Furthermore, GluA1-HT vesicles in Zone 1 imaged 10 min after the cessation of sLTP stimulation exhibited similar motion behaviors to those imaged immediately after the cessation of sLTP (Figure 2—figure supplement 3A, Active transport bar graph and Diffusion histogram), suggesting that vesicles are confined.

To further test whether vesicles are confined near the site of synaptic activity, we used HMM-Bayes to determine the probabilities that vesicles switch between active transport and diffusion in each zone during sLTP. sLTP increases the probability that multi-state GluA1-HT vesicles undergoing active transport in Zone 1 switch to and stay in a diffusive state (Figure 2—figure supplement 4, sLTP Zone 1, k_D-AT vs k_AT-D, and k_D-D) in a manner dependent on the presence of MNI (Figure 2—figure supplement 4, No MNI, k_D-AT vs k_AT-D, and k_D-D). By contrast, multi-state vesicles in Zone 2 and Zone 3 have similar probabilities of switching between active transport and diffusion after sLTP (Figure 2—figure supplement 4, sLTP Zone 2+3, k_D-AT vs k_AT-D). These observations demonstrate that multi-state GluA1-HT vesicles proximal to the site of stimulation have low probabilities of being transported away from the site of stimulation. Combined, these findings demonstrate that sLTP results in the confinement of GluA1-HT vesicles near the site of structural plasticity.

Confinement of GluA1 vesicles during synaptic activity is mediated by F-actin-induced molecular crowding in the dendritic shaft

Having demonstrated that sLTP results in the confinement of GluA1-HT vesicles in the dendritic shaft near the site of structural plasticity, we sought to determine the molecular mechanisms involved in disrupting vesicle motion. We hypothesized that F-actin in the dendritic shaft might be involved because stimulating neuronal activity leads to elevated intracellular calcium levels and the activation of calcium signaling pathways that trigger the polymerization of actin (Okamoto et al., 2004; Okamoto et al., 2007; Okamoto et al., 2009). Moreover, recent studies have reported the rearrangement of F-actin networks in the dendritic shaft during neuronal activity (Schätzle et al., 2018; Lavoie-Cardinal et al., 2020), and found that F-actin networks can reposition lysosomes in neurites (Katrukha et al., 2017; van Bommel et al., 2019). Importantly, F-actin networks formed in response to sLTP are persistent (Okamoto et al., 2004), and therefore could be a mechanism to confine GluA1-HT vesicles even after the cessation of stimulation.

To determine whether actin polymerization occurs in the dendritic shaft of cultured rat hippocampal neurons during synaptic activity, we tested whether cLTP would lead to the redistribution of F-tractin (tractin), an actin binding peptide that is used as a marker for F-actin (Schell et al., 2001). Prior to cLTP, tractin is diffusely distributed in the dendritic shaft and concentrated in spines (Figure 3A, Control). During cLTP induction, tractin in the dendritic shaft redistributes into a network of filaments (Figure 3A, cLTP). The combined length of these tractin filaments in the dendritic shaft is significantly greater during cLTP (Figure 3A, Tractin length, and Figure 3—figure supplement 1), suggesting that cLTP stimulates actin polymerization in the dendritic shaft. To eliminate the possibility that changes in the distribution of tractin are due to morphological changes in the dendrite, neurons were also transduced with a plasmid expressing tdTomato, which did not dramatically redistribute during cLTP (Figure 3—figure supplement 2A).

Figure 3 with 9 supplements see all

Download asset Open asset

Reduced motion of GluA1-HT vesicles during synaptic activity is mediated by actin polymerization in the dendritic shaft.

(A) Images: representative Airyscan images of F-tractin-mNeongreen (Tractin) in a dendrite before treatment (Control) and during cLTP (cLTP). Scale bar, 5 μm. Graph: combined length of tractin filaments (Tractin length) before treatment and during cLTP. **p=0.0039 by Wilcoxon matched pairs test. Each dot represents the tractin length in an imaged region of a dendrite. n=9 dendrite regions (each from one neuronal culture). (B) Same as (A), but in the presence of Latrunculin A (LatA). Scale bar, 5 μm. n=8 dendrite regions. (C) Representative epifluorescence images of tractin immediately before sLTP (Pre-sLTP) and after sLTP (Post-sLTP). Blue dot represents the site of uncaging. Blue outline denotes area with increased tractin signal after sLTP stimulation. Green inset: magnified image of tractin around the uncaging site. Scale bars, 10 μm. (D) GluA1-HT vesicle trajectories immediately before sLTP (Pre-sLTP) and after sLTP (Post-sLTP) inside (IN) and outside (OUT) the region where there was increased actin polymerization after sLTP stimulation. Green inset: magnified image of trajectories around the uncaging site. For video, see Figure 3—video 1. (E) Line graphs of fractions of GluA1-HT vesicles exhibiting active transport inside (IN) and outside (OUT) regions of actin polymerization pre- and post-sLTP. **p=0.0039 by Wilcoxon matched pairs test. Each dot represents the fraction of vesicles exhibiting active transport inside or outside the region with actin polymerization for a single sLTP stimulation experiment. n=9 sLTP stimulation experiments (each experiment targets one spine on one neuron) where GluA1-HT vesicle motion in the dendrite is captured immediately before and after sLTP stimulation. (F) Distributions of diffusion coefficients for GluA1-HT vesicles in regions with actin polymerization (IN; blue) versus regions without actin polymerization (OUT; gray) after sLTP stimulation. Lines represent the probability density function of each histogram estimated by kernel density estimation (KDE). **p=0.0053 by Kolmogorov-Smirnov test. n=59–73 GluA1-HT vesicle trajectories pooled together from nine sLTP stimulation experiments. (G) Same as (E), but in the presence of LatA. IN region defined as the 30 μm region flanking the uncaging site (the average length of dendrite where actin polymerization occurs after sLTP; Figure 3—figure supplement 3A), as we do not detect actin polymerization in the presence of LatA. n=7 sLTP stimulation experiments. (H) Distributions of diffusion coefficients for GluA1-HT vesicles in regions with actin polymerization (IN; blue) versus GluA1-HT vesicles in similar sized regions in the presence of LatA (LatA IN; red). **p=0.0013 by Kolmogorov-Smirnov test. n=59–67 GluA1-HT vesicle trajectories pooled together from seven to nine sLTP stimulation experiments for each condition. (**I–J**) Bar graphs of adjusted vesicle counts inside (IN) or outside (OUT) regions with actin polymerization after sLTP (I) or sLTP in the presence of LatA (J). Error bars represent standard deviation. *p=0.0111 by Mann-Whitney test. n=7–9 sLTP stimulation experiments for each condition.

Figure 3—source data 1 Related to Figure 3.: https://cdn.elifesciences.org/articles/80622/elife-80622-fig3-data1-v1.xlsx
Download elife-80622-fig3-data1-v1.xlsx

To confirm that the change in tractin distribution during cLTP induction is the result of actin polymerization, we imaged tractin during cLTP in dendrites that were also treated with Latrunculin A (LatA), an inhibitor of actin polymerization (Figure 3B and Figure 3—figure supplement 2B). Treating neurons with LatA not only prevents redistribution of tractin during cLTP, but also reduces the intensity of tractin signal in dendritic spines (Figure 3B, cLTP +LatA), demonstrating that actin polymerization leads to the redistribution of tractin during cLTP. Because the expression of actin-binding peptides may result in artificial F-actin structures (Melak et al., 2017), we also labeled F-actin with phalloidin after cLTP and imaged using STED microscopy. cLTP resulted in greater phalloidin labeling in dendritic shafts, while LatA treatment during cLTP decreased phalloidin labeling (Figure 3—figure supplement 2C), recapitulating our finding that cLTP induces actin polymerization in the dendritic shaft.

We next sought to determine whether sLTP-mediated changes in local actin networks play a role in positioning vesicles near sites of stimulation. We observed a significant increase in tractin fluorescence (MFI) in spines stimulated with sLTP and in the dendritic shaft proximal to these spines, reflecting increased actin polymerization at these locations (Figure 3C, blue outline). Tractin signal increases in an approximately 30 μm longitudinal section along the length of the dendritic shaft surrounding the sLTP-stimulated spine (Figure 3—figure supplement 3A). The increase in tractin MFI during sLTP is dependent on actin polymerization and is not an artifact of photostimulation (Figure 3—figure supplement 3B).

Having found that sLTP increases actin polymerization in the dendritic shaft proximal to the uncaging site, we tested whether sLTP-mediated actin polymerization confined GluA1-HT vesicles. By tracking GluA1-HT vesicles after sLTP (Figure 3D and Figure 3—video 1), we find that sLTP significantly reduces the fraction of GluA1-HT vesicles undergoing active transport, as well as the diffusion coefficient of GluA1-HT vesicles, inside but not outside regions of the dendritic shaft with actin polymerization (Figure 3E–F). Moreover, multi-state GluA1-HT vesicles undergoing active transport inside, but not outside, regions of actin polymerization during sLTP have an increased probability of switching to diffusion (Figure 3—figure supplement 4). By contrast, sLTP stimulation has no effect on the motion of vesicles in cultures with no MNI (Figure 3—figure supplement 3C–D). Importantly, the effect of sLTP on GluA1-HT vesicle motion is disrupted by LatA treatment (Figure 3G–H and Figure 3—figure supplement 4). Similarly, LatA prevented cLTP-mediated changes in GluA1-HT vesicle mobility, demonstrating that cLTP-induced actin polymerization results in the confinement of GluA1-HT vesicles as well (Figure 3—figure supplement 5 and Figure 1—figure supplement 11).

These results demonstrate that sLTP-mediated actin polymerization in the dendritic shaft confines GluA1-HT vesicles near the site of stimulation, but it is unclear whether confinement actually generates an increased number of vesicles near these sites. After adjusting the number of vesicles for photobleaching (Materials and methods), we find a significant increase in the number of GluA1-HT vesicles inside, but not outside, regions of actin polymerization after sLTP (Figure 3I and Figure 3—figure supplement 3E). Moreover, the increase in GluA1-HT vesicles is blocked by the addition of LatA (Figure 3J). Combined, these observations demonstrate that neurons use actin polymerization as a mechanism to confine and increase the number of vesicles near the sites of synaptic activity.

Having established actin polymerization as the mechanism that mediates stimulation-dependent GluA1-HT vesicle confinement, we sought to determine the mechanism by which F-actin perturbed vesicle motion. AMPARs interact with myosins Va, Vb, and VI (Correia et al., 2008; Wang et al., 2008; Nash et al., 2010; EstevesdaSilva et al., 2015), which are involved in the calcium-dependent, short-range recruitment of AMPARs in endosomes to and from dendritic spines. To test if interactions between GluA1 and myosin V and/or VI anchor GluA1 vesicles to F-actin during neuronal activity, we inhibited myosin Va, Vb, and VI by expressing dominant-negative c-terminal domains of these proteins, and by using a pharmacological inhibitor cocktail (MI), during cLTP stimulation (Figure 3—figure supplement 6A–C, Figure 3—video 2 and Figure 3—video 3). Inhibition of myosin did not alter the fractions of GluA1-HT vesicles exhibiting active transport, or the diffusion coefficients of GluA1-HT vesicles, either under unstimulated conditions or during cLTP (Figure 3—figure supplement 6B–C). Likewise, acute pharmacological inhibition of myosin did not affect the fraction of GluA1-HT vesicles exhibiting active transport, or the diffusion coefficient of GluA1-HT vesicles, after sLTP stimulation (Figure 3—figure supplement 6D). Based on these observations, we conclude that F-actin disrupts GluA1-HT vesicle motion in a manner independent of myosin activity.

Previous studies have demonstrated that F-actin can constrain the motion of lysosomes (van Bommel et al., 2019), leading us to speculate that actin polymerization itself could block the motion of GluA1-HT vesicles. Treatment of neurons with Jasplakinolide (Jsp), an F-actin stabilizer that promotes actin polymerization, is sufficient to inhibit active transport and reduce the rate of diffusion of GluA1-HT vesicles (Figure 4A and Figure 4B, left histogram). Furthermore, the Jsp-mediated reduction in GluA1-HT vesicle motion is not altered by pharmacological inhibition of myosin, indicating that F-actin itself plays a role in the reduced motion of GluA1-HT vesicles (Figure 4A and Figure 4B, right histogram).

Figure 4 with 1 supplement see all

Download asset Open asset

F-actin-induced molecular crowding inhibits the motion of particles near the sites of synaptic activity.

(A) Fractions of GluA1-HT vesicles exhibiting active transport in dendritic shafts without treatment (Con) versus during treatment with Jasplakinolide (Jsp) or Jsp with pharmacological inhibition of myosin (Jsp +MI). Bars represent mean and standard deviation. Con vs Jsp, ****p<0.0001. Significance was determined by Mann-Whitney test. n=9–12 timelapse imaging sequences (each timelapse captures the motion of GluA1-HT vesicles in one region of dendrite in one neuronal culture) for each condition. (B) Left: distributions of diffusion coefficients of GluA1-HT vesicles in dendritic shafts without treatment (Control; gray) versus during treatment with Jsp (Jsp; blue). Lines represent the probability density function of each histogram estimated by kernel density estimation (KDE). Control vs Jsp, ****p<0.0001 by Kolmogorov-Smirnov test. Right: distributions of diffusion coefficients of GluA1-HT vesicles in dendritic shafts during treatment with Jsp (Jsp; blue) versus Jsp with pharmacological inhibition of myosin (Jsp +MI; red). n=60–227 GluA1-HT vesicle trajectories pooled from 9 to 12 timelapse imaging sequences for each condition. (C) Schematic: GEM-HT-ST rheological probe (top). GEM-HT-ST was created by fusing HaloTag (HT) and SNAP-tag (ST) to the PfV protein from *Pyroccocus furiosus* (bottom; Delarue et al., 2018). When expressed, PfV-HT-ST fusion proteins self-assemble into a 40 nm encapsulin scaffold that can be labeled with either JF₅₄₉-HTL or JF₅₄₉-SNAP-tag ligand (JF₅₄₉-STL). Image: trajectories of GEM-HT-ST labeled with JF₅₄₉-HTL. Scale bar, 10 μm. For video, see Figure 4—video 1. (D) Distributions of diffusion coefficients of GEM-HT-ST after chemical stimulation. Control (gray) vs cLTP (blue), **p=0.0047; Control (gray) vs Jsp (blue), **p=0.0011. Significance was determined by Kolmogorov-Smirnov test. n=57–94 GEM-HT-ST trajectories pooled together from six to nine timelapse imaging sequences for each condition. (E) Distributions of diffusion coefficients of GEM-HT-ST after sLTP in regions where actin polymerization occurred versus controls. IN region was determined by isolating a 30 μm region flanking the uncaging site (the average length of dendrite where actin polymerization occurs after sLTP; Figure 3—figure supplement 3A). sLTP IN (blue) vs sLTP OUT (gray), **p=0.0083; sLTP IN (blue) vs No MNI IN (gray), *p=0.0446. n=44–59 GEM-HT-ST trajectories pooled together from six to nine sLTP stimulation experiments (each experiment targets one spine on one neuron) where GEM-HT-ST motion in the dendrite is captured immediately after sLTP stimulation.

Figure 4—source data 1 Related to Figure 4.: https://cdn.elifesciences.org/articles/80622/elife-80622-fig4-data1-v1.xlsx
Download elife-80622-fig4-data1-v1.xlsx

Previous studies have found that actin can induce molecular crowding in biological systems such as neuronal axons and prevent active transport of vesicles and organelles (Sood et al., 2018). To test if actin polymerization could disrupt vesicle motion by changing the rheological properties of the dendritic cytoplasm, we evaluated whether cLTP and sLTP can alter the diffusion of a rheological probe consisting of a genetically encoded multimeric nanoparticle (GEM; Delarue et al., 2018; Figure 4C). GEM is based on the encapsulin protein PfV of Pyroccocus furiosus (Figure 4C, schematic), which self-assembles into an icosahedral scaffold of 120 monomers whose size is more similar to GluA1-HT vesicles than are soluble fluorescent proteins. When fused to both HaloTag and the self-labeling SNAP-tag (ST), the particle arising from expression of GEM-HT-ST can be labeled with Janelia Fluor dye ligands and tracked in dendrites (Figure 4C, image, and Figure 4—video 1).

We find that GEM-HT-ST diffusion is significantly reduced during cLTP induction (Figure 4D, cLTP), and this reduction is dependent on actin polymerization as LatA prevents the decrease in diffusion coefficient (Figure 4D, cLTP +LatA). Moreover, actin polymerization stimulated by Jsp is sufficient to reduce the motion of GEM-HT-ST (Figure 4D, Jsp). sLTP also results in a reduction in the rate of GEM-HT-ST diffusion in regions of dendritic shafts where actin polymerization occurs (Figure 4E, sLTP IN vs sLTP OUT) in a manner that is dependent on the presence of MNI (Figure 4E, No MNI IN) and actin polymerization (Figure 4E, sLTP +LatA IN). These experiments demonstrate that synaptic activity changes the rheological properties of the dendritic shaft by stimulating actin polymerization. Combined with our findings that triggering actin polymerization is sufficient to disrupt GluA1-HT vesicle motion (via Jsp treatment) and that myosin inhibition did not alter activity-mediated changes in motion, these observations are consistent with the hypothesis that actin polymerization confines GluA1-HT vesicles by altering the properties of the dendritic cytoplasm independent of direct interactions between GluA1 and myosin.

Local increase in GluA1 exocytosis triggered by synaptic activity is dependent on actin-mediated GluA1 vesicle confinement and myosin activity

Although we have shown that actin polymerization can concentrate GluA1 vesicles near the sites of synaptic activity, it is unclear whether this mechanism contributes to increased trafficking of GluA1 to synapses during activity – whether positioning vesicles near the sites of activity also results in increased exocytosis of GluA1-HT at these sites. To study the exocytic rates of endogenous GluA1, we fused super ecliptic pHluorin (SEP) to HaloTag and knocked this tandem reporter into the NTD of GluA1 such that the tag is exposed to the low pH lumen of vesicles during transport (Figure 5A and Figure 5—figure supplements 1–2). Intracellular GluA1-HT-SEP has low fluorescence until exocytosis, at which point it is exposed to the neutral pH extracellular medium and exhibits strong fluorescence (Figure 5A, diagram). During exocytosis, GluA1-HT-SEP released from vesicles is temporarily confined at the sites of exocytosis and appears as bright puncta under fluorescence microscopy (Figure 5A, images on right, and Figure 5—video 1). GluA1-HT-SEP exocytic events occur at a low rate prior to stimulation, but increase dramatically during cLTP induction (Figure 5B, cLTP, and Figure 5—video 2), consistent with previous findings (Kopec et al., 2006). The increase in GluA1-HT-SEP exocytosis is dependent on actin polymerization, as LatA reduces the rate of exocytosis during cLTP induction (Figure 5B, cLTP +LatA). Acute pharmacological inhibition of myosin also disrupts cLTP-stimulated GluA1-HT-SEP exocytosis (Figure 5B, cLTP +MI), suggesting that while myosin does not play a role in disrupting transport of GluA1-HT vesicles, it plays a role in regulating exocytosis of GluA1-HT-SEP. These observations demonstrate that actin polymerization and myosin mediate GluA1-HT-SEP exocytosis during cLTP.

Figure 5 with 6 supplements see all

Download asset Open asset

Increased GluA1-HT-SEP exocytosis triggered by synaptic activity is dependent on actin polymerization and myosin activity.

(A) Tagging endogenous GluA1 with a HaloTag-pHluorin tandem fusion reporter (HT-SEP) to track GluA1 exocytosis events. Diagram: endogenous GluA1 tagged with HT-SEP (GluA1-HT-SEP, schematic of fusion on right) has low SEP fluorescence inside vesicles due to the low pH of the vesicle lumen. When a GluA1 exocytosis event occurs, SEP will be exposed to the neutral pH of the extracellular medium, resulting in increased fluorescence. Because receptors are temporarily spatially restricted during exocytosis, a spot with fluorescence can be observed during the event. Images: timelapse of two GluA1-HT-SEP exocytosis events (yellow circles). Scale bar, 5 μm. For video, see Figure 5—video 1. (B) Bar graph of GluA1-HT-SEP exocytosis in response to cLTP in the absence or presence of Latrunculin A (LatA) or pharmacological myosin inhibition (MI). Bars represent mean events per μm per min and standard deviation. Con vs cLTP, ****p<0.0001; cLTP vs cLTP +LatA, ****p<0.0001; cLTP vs cLTP +MI, ****p<0.0001. Significance was determined by Mann-Whitney test. n=12 timelapse imaging sequences (each timelapse captures GluA1-HT-SEP exocytosis in one region of dendrite in one neuronal culture) for each condition. (C) Line graphs of GluA1-HT-SEP exocytic events before and after sLTP (sLTP), sLTP in the presence of LatA (sLTP +LatA), and sLTP in the presence of MI (sLTP +MI). Proximal: GluA1-HT-SEP exocytic events within the 30 μm region surrounding the site of uncaging (i.e. the average length of dendrite where actin polymerization occurs near the site of uncaging after sLTP; see Figure 3—figure supplement 3A). sLTP, ***p=0.0005; sLTP +LatA, *p=0.0312; sLTP +MI, *p=0.0312. Distal: GluA1-HT-SEP exocytic events outside of the 30 μm region surrounding the site of uncaging. sLTP, *p=0.0430. Significance was determined by Wilcoxon matched pairs test. Each dot represents the number of exocytic events in the indicated region. n=6–12 sLTP stimulation experiments (each experiment targets one spine on one neuron) where GluA1-HT-SEP exocytosis in the dendrite is captured immediately before and after sLTP stimulation.

Figure 5—source data 1 Related to Figure 5.: https://cdn.elifesciences.org/articles/80622/elife-80622-fig5-data1-v1.xlsx
Download elife-80622-fig5-data1-v1.xlsx

We next tested whether sLTP results in local increases in GluA1-HT-SEP exocytosis, and if the increase in exocytosis is spatially correlated with, and dependent on, actin polymerization (Figure 5C and Figure 5—video 3). In the absence of a marker for F-actin (due to overlapping fluorescence signals between tractin and SEP), we defined the area proximal to the site of stimulation – the 30 µm region surrounding the uncaging site – as the region of actin polymerization based on our previous observations (Figure 3—figure supplement 3A). sLTP stimulation increases GluA1-HT-SEP exocytosis events to a much greater extent proximal than distal to the site of uncaging (Figure 5C, sLTP). Similar to their effect on cLTP-mediated GluA1-HT-SEP exocytosis, LatA treatment and acute myosin inhibition both partially block sLTP-mediated increases in GluA1-HT-SEP exocytosis (Figure 5C, sLTP +LatA and sLTP +MI). When MNI is removed from the media, there is no increase in exocytic events after sLTP (Figure 5—figure supplement 3). We conclude that the accumulation of GluA1-HT vesicles near sites of sLTP is spatially correlated with increased exocytosis of GluA1-HT-SEP, and that local disruption of GluA1-HT vesicle motion and the increase in GluA1-HT-SEP exocytosis are both dependent on actin polymerization in the dendritic shaft.

To determine if the concentration of GluA1-HT-SEP vesicles near the site of synaptic activity contributes to increased trafficking of GluA1-HT-SEP to the cell surface, we sparsely labeled GluA1-HT-SEP vesicles with JF dye ligands and simultaneously tracked vesicle motion and exocytosis (Figure 6A and Figure 6—video 1). We sought to determine whether GluA1-HT-SEP vesicles destined for exocytosis (i.e. pre-exocytosis vesicles) are drawn from a local source or from distal loci via long-range active transport immediately prior to exocytosis. If pre-exocytosis GluA1-HT-SEP vesicles are drawn from local sources, they should travel relatively short net distances to the sites of exocytosis. To test this hypothesis, we measured the radius of confinement (the area in which a trajectory is confined) for the trajectories of pre-exocytosis GluA1-HT-SEP vesicles in unstimulated cells (Figure 6B, top bar graph). The trajectories of GluA1-HT-SEP vesicles that undergo exocytosis (Pre-exocytosis) have smaller radii of confinement when compared to vesicles that do not exocytose (Non-exocytosis), demonstrating that GluA1-HT-SEP vesicles are not imported from distal loci immediately prior to exocytosis.

Figure 6 with 3 supplements see all

Download asset Open asset

GluA1-HT-SEP vesicles are drawn from a local pool in the dendritic shaft prior to exocytosis and exhibit lateral motion that is dependent on actin polymerization and myosin in the shaft during synaptic activity.

(A) Left: representative maximum intensity projection (MIP) of GluA1-HT-SEP exocytosis over time after sLTP (i.e. the brightest pixels from each image in a timelapse compressed into a single image). Yellow circles indicate exocytic events. Blue dot indicates site of uncaging. Right: GluA1-HT-SEP vesicle trajectories overlaid on the MIP of exocytosis over time. Blue trajectories are GluA1-HT-SEP vesicles that undergo exocytosis (Pre-exocytosis). Gray trajectories are GluA1-HT-SEP vesicles with no detected exocytosis (Non-exocytosis) during the span of imaging. Scale bar, 5 μm. For video. See Figure 6—video 1. (B) Top: mean radius of confinement for the trajectories of GluA1-HT-SEP vesicles that do not undergo exocytosis (Non-exocytosis) and that do undergo exocytosis (Pre-exocytosis) under unstimulated conditions. Error bars represent standard deviation. ****p<0.0001 by Kolmogorov-Smirnov test. n=65–176 GluA1-HT-SEP vesicle trajectories pooled from five timelapses imaging experiments (each timelapse captures GluA1-HT-SEP motion and exocytosis in one region of dendrite in one neuronal culture). Bottom: bar graph of motion types (Figure 6—figure supplement 1) for non-exocytosis GluA1-HT-SEP vesicles (gray bars) and pre-exocytosis GluA1-HT-SEP vesicles (black dots) after sLTP. Only trajectories in regions where actin polymerization occurred were used (see Figure 3—figure supplement 3A). No MNI vs sLTP, **p=0.0018; sLTP vs sLTP +LatA, ***p=0.0003; sLTP vs sLTP +MI, **p=0.0017. Significance was determined by Mann-Whitney test. n=9–12 sLTP stimulation experiments (each experiment targets one spine on one neuron) where GluA1-HT-SEP motion and exocytosis in the dendrite are captured immediately after sLTP stimulation. (C) Diagram of directional bias analysis. The directional bias of a trajectory can be determined by calculating the angle, theta (Θ), created between the line from the center of mass of a trajectory (CM) to the nearest point on the dendrite path and the line from the CM to the furthest point from the CM. Theta of 90^o indicates longitudinal movement while theta of 0^o indicates lateral movement. Images: representative trajectories of GluA1-HT vesicles traveling longitudinally under control conditions (Control; top) or laterally after cLTP (cLTP; bottom). (D) Directional bias for GluA1-HT-SEP vesicles prior to exocytosis (pre-exocytosis) after sLTP. Top: distribution of theta for vesicles after sLTP. sLTP (blue) vs No MNI (gray), **p=0.0067. Middle: distribution of theta for vesicles after sLTP in the presence of an actin polymerization inhibitor (LatA). sLTP +LatA (blue) vs sLTP (gray), **p=0.0032. Bottom: distribution of theta for vesicles after sLTP in the presence of acute myosin inhibition (MI). sLTP +MI (blue) vs sLTP (gray), **p=0.0078. Lines represent the probability density function of each histogram estimated by kernel density estimation (KDE). Significance was determined by Kolmogorov-Smirnov test. n=31–50 GluA1-HT-SEP trajectories pooled from 9 to 12 sLTP stimulation experiments for each condition.

Figure 6—source data 1 Related to Figure 6.: https://cdn.elifesciences.org/articles/80622/elife-80622-fig6-data1-v1.xlsx
Download elife-80622-fig6-data1-v1.xlsx

Based on the observations that pre-exocytosis GluA1-HT-SEP vesicles have small search spaces and that stimulation reduces active transport for GluA1-HT vesicles (Figure 1D–E and Figure 2B–C), we anticipated that stimulation would also reduce active transport for pre-exocytosis GluA1-HT-SEP vesicles, and that these vesicles diffuse over short distances to the sites of exocytosis. To test this idea, we used HMM-Bayes to infer the motion of pre-exocytosis GluA1-HT-SEP vesicles after inducing structural plasticity with sLTP stimulation. Pre-exocytosis GluA1-HT-SEP vesicles exhibit two or more motion states, where the final state before exocytosis is immobility (i.e. vesicles are immobilized by docking immediately prior to exocytosis; Figure 6—figure supplement 1A). Thus, we characterized the motion states of vesicles prior to immobility (docking). Similar to GluA1-HT vesicles, non-exocytosis GluA1-HT-SEP vesicles (i.e. vesicles that do not exocytose) exhibit reduced active transport in response to sLTP (Figure 6B, bottom bar graph, No MNI vs sLTP, and Figure 6—figure supplement 1B, Non-exocytosis). Surprisingly, sLTP increased the fraction of pre-exocytosis GluA1-HT-SEP vesicles exhibiting active transport (Figure 6B, bottom line graph, No MNI vs sLTP, and Figure 6—figure supplement 1B, Pre-exocytosis). Similarly, cLTP also increased active transport of pre-exocytosis GluA1-HT-SEP vesicles (Figure 6—figure supplement 1C). These results show that synaptic activity stimulates the active transport of GluA1-HT-SEP vesicles to exocytic sites, even though the net distances they travel are short.

As active transport delivers vesicular cargo much faster than diffusion, this finding is consistent with the model that AMPARs need to be rapidly trafficked to exocytic sites during synaptic activity in order to maintain a rapidly accessible membrane-bound reservoir of receptors (Huganir and Nicoll, 2013). Nevertheless, it is striking that synaptic activity increases the active transport of pre-exocytosis GluA1-HT-SEP vesicles while simultaneously confining the motion of non-exocytosis GluA1-HT-SEP vesicles. AMPAR-containing endosomes are primarily trafficked along to the length of dendrites by microtubule-based motors (EstevesdaSilva et al., 2015; Setou et al., 2002; Hoerndli et al., 2013), but can also be recruited by myosin Va and Vb to the sites of exocytosis during LTP induction (Correia et al., 2008; Wang et al., 2008). Thus, we hypothesized that actin polymerization in the dendritic shaft plays a dual role during synaptic activity by: (1) disrupting microtubule-based transport of GluA1 vesicles near synaptic activity (increasing the concentration of GluA1 vesicles); and (2) acting as a substrate for the myosin-based transport of a minority of vesicles to exocytic sites. This may explain the different motion states we observe between pre- and non-exocytosis GluA1-HT-SEP vesicles during stimulation. Interestingly, we primarily observe GluA1-HT vesicles undergoing transport parallel to the length of the dendritic shaft (longitudinal motion) prior to stimulation (Figure 1D, control and Figure 2B, Pre-sLTP). After stimulation, we find GluA1-HT vesicles that exhibit motion perpendicular to the length of the dendritic shaft (lateral motion; Figure 1D, cLTP and Figure 2B, Post-sLTP). Consequently, we asked if vesicles that changed their directional bias in response to stimulation were in fact pre-exocytosis vesicles being transported to their exocytic sites by myosin.

To examine this possibility, we first measured the directional bias, described by the angle theta, of pre-exocytosis GluA1-HT-SEP vesicle trajectories in response to sLTP-stimulated synaptic activity (Figure 6C). Theta of ~90^o indicates that the vesicle is moving parallel to the dendritic shaft (longitudinal motion; Figure 6C, top trajectory), whereas theta of ~0^o indicates the vesicle is moving perpendicular to the dendritic shaft (lateral motion; Figure 6C, bottom trajectory). When we examine the directional bias for non-exocytosis GluA1-HT-SEP vesicles in the absence of sLTP stimulation, we find a strong bias for longitudinal motion (Figure 6—figure supplement 2A, Non-exocytosis vesicles). By contrast, pre-exocytosis GluA1-HT-SEP vesicles do not exhibit strong biases for either longitudinal or lateral motion in the absence of stimulation (Figure 6D, top histogram, No MNI, and Figure 6—figure supplement 2A, Pre-exocytosis vesicles), indicating that vesicles travel in random directions to exocytic sites in the absence of synaptic activity. However, when synaptic activity is induced by sLTP, pre-exocytosis GluA1-HT-SEP vesicles have a strong bias for lateral motion (Figure 6D, top histogram, sLTP). Likewise, cLTP induction also leads to increased lateral motion for pre-exocytosis GluA1-HT-SEP vesicles (Figure 6—figure supplement 2B, Control vs cLTP). These results show vesicles move laterally to exocytic sites in response to synaptic activity, which we speculate is due to vesicles moving from the center to the periphery of the dendritic shaft.

We then sought to determine whether increased lateral motion during synaptic activity is driven by myosin. We first examined the motion states of pre-exocytosis GluA1-HT-SEP vesicles that move laterally during stimulation and find most exhibit active transport (Figure 6—figure supplement 2C). This result suggests that GluA1-HT-SEP vesicles move by motor-based transport to exocytic sites. Moreover, when actin polymerization or myosin activity are inhibited, the fraction of pre-exocytosis GluA1-HT-SEP vesicles exhibiting active transport is significantly reduced (Figure 6B, bottom line graph, sLTP +LatA and sLTP +MI). Next, we tested the effect of inhibiting actin polymerization on directional bias and find that LatA strongly reduces the lateral motion of pre-exocytosis GluA1-HT-SEP vesicles in response to sLTP (Figure 6D, middle histogram, sLTP +LatA). Similarly, pharmacological inhibition of myosin also blocked lateral motion in response to sLTP (Figure 6D, bottom histogram, sLTP +MI). Inhibition of either actin polymerization or myosin activity also prevents increased lateral motion of pre-exocytosis vesicles in response to cLTP (Figure 6—figure supplement 2B, cLTP vs cLTP +LatA and cLTP +MI). Together, these findings demonstrate that pre-exocytosis GluA1-HT-SEP vesicles are transported laterally by myosin to the sites of exocytosis in response to synaptic activity.

Lastly, we sought to confirm our finding that actin polymerization itself is sufficient to block longitudinal motion. We examined theta for GluA1-HT vesicles in response to stimulation and find a decrease in longitudinal motion and increase in lateral motion (Figure 6—figure supplement 2D, sLTP and cLTP), similar to what we observe for pre-exocytosis vesicles. However, when GluA1-HT vesicles are treated with LatA during stimulation, we observe a decrease in lateral motion and a strong increase in longitudinal motion (Figure 6—figure supplement 2E, sLTP +LatA and cLTP +LatA). By contrast, inhibition of myosin during stimulation reduces lateral motion but does not increase longitudinal motion (Figure 6—figure supplement 2F, sLTP +MI and cLTP +MI, and Figure 6—figure supplement 2G). Moreover, inducing actin polymerization with Jasplakinolide (Jsp) while blocking myosin activity reduces longitudinal motion without increasing lateral motion (Figure 6—figure supplement 2H, Jsp +MI). These observations support our conclusion that F-actin is necessary and sufficient to confine GluA1 vesicle motion near sites of synaptic activity. Nevertheless, myosin also promotes the surface expression of GluA1 by mediating its active transport to the sites of exocytosis.

Discussion

In this study, we have developed a novel method to identify, track and characterize the motion of vesicles containing GluA1, enabling us to better understand how AMPARs are delivered specifically to sites undergoing plasticity. We use homology-independent targeted integration (HITI) to tag endogenous GluA1 with HaloTag (GluA1-HT) and then a block-and-chase labeling protocol with Janelia Fluor (JF) dye ligands to achieve a sparse labeling density suitable for the detection of GluA1-HT vesicles. We then utilize single-particle tracking (SPT) followed by hidden Markov modeling with Bayesian model selection (HMM-Bayes) to describe the motion of GluA1-HT vesicles during chemical and structural LTP (cLTP and sLTP). Using this strategy, we find that GluA1-HT vesicles become confined by actin polymerization in the dendritic shaft proximal to sites of stimulation, resulting in an increased intracellular reservoir of GluA1-HT near these sites. Using a pHluorin-HaloTag fusion with GluA1 (GluA1-HT-SEP), we examine how local vesicular reservoirs of GluA1 contribute to GluA1 exocytosis. We find that pre-exocytosis GluA1-HT-SEP vesicles undergo short-range transport perpendicular to the length of the dendritic shaft near sites of stimulation in a manner dependent on both actin polymerization and myosin activity.

Based on these findings, we propose a new model in which neurons utilize actin polymerization in the dendritic shaft to specify the location to which AMPARs are delivered during synaptic activity. First, actin polymerization occurs in the dendritic shaft proximal to the site of synaptic activity, resulting in molecular crowding in the dendritic cytoplasm at this location (Figure 7A–B). The increased crowding inhibits the longitudinal motion of GluA1 vesicles, concentrating intracellular GluA1 near the site of synaptic activity (Figure 7A–B). F-actin then acts as a substrate for myosin Va and Vb – activated by the influx of calcium (Correia et al., 2008; Wang et al., 2008) – to recruit GluA1 vesicles to the dendrite membrane (Figure 7C). Here, GluA1 vesicles undergo exocytosis, increasing the amount of surface bound GluA1 that can then diffuse into synapses (Figure 7D).

Figure 7

Download asset Open asset

Actin polymerization in the dendritic shaft proximal to the site of synaptic activity promotes GluA1 exocytosis by increasing the local pool of GluA1 vesicles and facilitating myosin-dependent transport to the dendrite periphery.

(A) Under unstimulated conditions, GluA1 vesicles are transported longitudinally along the dendritic shaft. (B) When synaptic activity is induced, actin polymerizes in the dendritic shaft near the site of activity. Actin polymerization disrupts the longitudinal motion of vesicles resulting in an increased pool of GluA1 vesicles near the site of activity. (C) Myosin V, which is inactive and sequestered in dendritic spines in unstimulated conditions, is activated by calcium influx during synaptic activity. Myosin subsequently translocates to the base of the dendritic spines. (D) Myosin V recruits GluA1 vesicles to the periphery resulting in increased exocytosis of GluA1.

Our labeling strategy has key advantages over previously described methods that enable us to image and track GluA1 vesicles. Primarily, we edit genomic copies of Gria1 to express GluA1-HT and thus avoid pitfalls associated with expressing tagged GluA1 from a plasmid – overexpressing GluA1 can lead to mislocalization and excess formation of calcium permeable GluA1 homomers (Diering and Huganir, 2018), altering conductance and neuronal activity. The low level of GluA1-HT expression driven by native Gria1 promoters is also advantageous because it enables us to achieve sparse particle labeling through a block-and-chase protocol without photobleaching. As photobleaching removes all signal from a designated area, particles are tracked as they travel from outside to inside this region. Consequently, data from photobleaching experiments may be biased towards fast moving particles (i.e. those with active transport) and may omit slowly diffusing particles. HITI can potentially introduce indels into target genes and knock out copies of Gria1 in some neurons (Suzuki et al., 2016), but we find a low frequency of indels around the insertion site of HaloTag (Figure 1—figure supplement 1D), and also that Gria1 expression levels in transfected neurons are similar to those in untransfected neurons (Figure 1—figure supplement 2B). These observations indicate that HITI-mediated editing of Gria1 does not often lead to unwanted mutations, especially those that knock out copies of Gria1. Nevertheless, tagging strategies with higher accuracy (e.g. vSlender; Nishiyama et al., 2017) could be viable alternatives to HITI.

Our observations of GluA1 vesicle trafficking differ somewhat from previous reports (EstevesdaSilva et al., 2015; Bowen et al., 2017; Hangen et al., 2018). Hangen et al., 2018 found that GluA1 vesicles undergoing active transport had decreased velocity and paused (i.e. temporarily lost active transport) more frequently during cLTP and photostimulation. By contrast, we find vesicles switch their motion state to diffusion and rarely switch back to active transport (i.e. vesicles stably lose active transport) in response to cLTP and sLTP. Vesicles also exhibit reduced diffusion coefficients after stimulation. Based on these differences, we conclude that GluA1 vesicles are confined, not paused, near sites of synaptic activity. These discrepancies are likely attributable to differences in our methodological approaches. Importantly, our labeling and analysis strategy enabled us to characterize the diffusion coefficient and state-switching probabilities of GluA1-HT vesicles, not just parameters associated with active transport. In addition, we use a specific MNI uncaging protocol to stimulate spine plasticity (not just calcium transients). We find that this protocol also induces actin polymerization in the dendritic shaft, which we demonstrate is the mechanism underlying vesicle confinement.

The precise mechanistic details of how actin polymerization is stimulated in the dendritic shaft and how it regulates GluA1 vesicles during synaptic activity remain to be determined. F-actin is found at the base of dendritic spines (Schätzle et al., 2018), and in the dendritic shaft in shaft synapses (van Bommel et al., 2019) and in periodic submembrane actin rings (Lavoie-Cardinal et al., 2020). Whether these F-actin networks remodel to regulate GluA1 vesicle motion during synaptic activity is not known. Furthermore, it is unclear if F-actin also coordinates the localization of other proteins and organelles near synaptic activity. For example, F-actin patches position lysosomes in shaft synapses to support AMPAR turnover (Goo et al., 2017; van Bommel et al., 2019), and thus could play a similar role near sites of synaptic activity. However, if actin-induced molecular crowding is generally obstructive to motion, additional mechanisms may be necessary to ensure only relevant particles localize to sites of synaptic activity. For example, myosin cargo adaptors that interact with GluA1 may further enhance the specificity of GluA1 transport (Correia et al., 2008; Hammer and Sellers, 2012), while other motors might enable GluA1-negative vesicles to bypass F-actin blockades (Ferro et al., 2019). Importantly, AMPARs are transported to multiple loci along a single dendrite, suggesting there must also be a mechanism to ensure not all GluA1 vesicles are trapped at a single synapse. One possibility is that there is a sufficient pool of GluA1 vesicles undergoing anterograde and retrograde transport to reach multiple synapses. Further investigation could clarify how actin polymerization is regulated in the dendritic shaft, and how F-actin filters specific cargo in response to synaptic activity.

Previous studies have found that AMPARs are primarily transported along the length of the dendrites on microtubules (EstevesdaSilva et al., 2015; Setou et al., 2002; Hoerndli et al., 2013; Hoerndli et al., 2015), but we and others find that short-ranged transport near the site of synaptic activity is mediated by Myosin Va/b (Correia et al., 2008; Wang et al., 2008). These observations suggest that GluA1 vesicles are transferred from microtubules to actin for local transport. The exact mechanism for this exchange is not known but our findings indicate that F-actin-induced crowding results in the detachment of GluA1 vesicles from microtubules prior to attachment and transport on F-actin. Thus, F-actin enhances local transport by acting as a substrate for myosin-based cargos, but also by creating a cytoplasmic environment that strongly disfavors microtubule-based transport (preventing transport away from the stimulated site). Such a mechanism could have important implications for how cargo in general is transferred since changes to the subcellular environment could alter modes of transport indirectly rather than through the interactions between motors, cargo, and cytoskeleton.

The precise control of synaptic protein trafficking is vital to synaptic transmission, and thus learning and memory. Nevertheless, many mechanisms regulating synaptic protein trafficking remain to be fully understood. Here, we identify a novel mechanism through which actin polymerization in the dendritic shaft can regulate the surface expression of GluA1 specifically at sites with stimulating inputs. Because F-actin can exert direct and indirect effects on a variety of particles in the dendritic cytoplasm, our findings raise interesting questions regarding whether actin polymerization is a general mechanism to coordinate the delivery of proteins during synaptic plasticity. Elucidating whether and how actin regulates the motion of proteins in the dendritic shaft during neuronal activity could help us better understand the cellular bases for learning and memory.

Materials and methods

Animal work was conducted according to the Institutional Animal Care and Use Committee guidelines of Janelia Research Campus (IACUC protocol #21–0206).

Localization error	1E-6
Deflation loops	3
Maximum number competitors	3
Maximum diffusion coefficient (µm²/s)	1	For GluA1-HT
Maximum diffusion coefficient (µm²/s)	3	For GEM

Reagent type (species) or resource	Designation	Source or reference	Identifiers	Additional information
Antibody	anti-AMPA receptor 1 (GluA1; Extracellular; Guinea pig polyclonal)	Alomone Labs	AGP-009	IF(1:250)
Antibody	anti-Chicken IgY, Alexa Fluor Plus 647 (Goat polyclonal)	Thermo Fisher Scientific	A-32933	IF(1:500)
Antibody	anti-Chicken IgY, Star Orange (Goat polyclonal)	abberior GmbH	STORANGE-1005–500 UG	IF(1:500)
Antibody	anti-Chicken IgY, Star Red (Goat polyclonal)	abberior GmbH	STRED-1005–500 UG	IF(1:500)
Antibody	anti-GluA1 (Mouse monoclonal)	Synaptic Systems	182 011	IF(1:500)
Antibody	anti-GluN1 (Mouse monoclonal)	Synaptic Systems	114 011	IF(1:1000)
Antibody	anti-Guinea Pig, Alexa Fluor 488 (Goat polyclonal)	Thermo Fisher Scientific	A-11073	IF(1:500)
Antibody	anti-Guinea Pig IgG, Alexa Fluor 555 (Goat polyclonal)	Thermo Fisher Scientific	A-21435	IF(1:500)
Antibody	anti-Guinea Pig IgG, Alexa Fluor 594 (Goat polyclonal)	Thermo Fisher Scientific	A-11076	IF(1:500)
Antibody	anti-Guinea Pig IgG, Alexa Fluor 647 (Goat polyclonal)	Thermo FIsher Scientific	A-21450	IF(1:500)
Antibody	anti-HaloTag (Rabbit polyclonal)	Promega	G9281	IF(1:500)
Antibody	anti-Homer1 (Chicken polyclonal)	Synaptic Systems	160 006	IF(1:1000)
Antibody	anti-Mouse IgG, Alexa Fluor 488 (Goat polyclonal)	Thermo Fisher Scientific	A-11001	IF(1:500)
Antibody	anti-Mouse IgG, Alexa Fluor 546 (Goat polyclonal)	Thermo Fisher Scientific	A-11030	IF(1:500)
Antibody	anti-Mouse IgG, Alexa Fluor 594 (Goat polyclonal)	Thermo Fisher Scientific	A-32742	IF(1:500)
Antibody	anti-Mouse IgG, Alexa Fluor 647 (Goat polyclonal)	Thermo Fisher Scientific	A-21236	IF(1:500)
Antibody	anti-Mouse IgG, Atto 647 N (Goat polyclonal)	Rockland Immunochemicals Inc.	610-156-121	IF(1:500)
Antibody	anti-PSD-95 (7E3-1B8; Mouse monoclonal)	Thermo Fisher Scientific	MA1-046	IF(1:1000)
Antibody	anti-PSD-95 (Guinea pig polyclonal)	Synaptic Systems	124 014	IF(1:2000)
Antibody	anti-Rabbit IgG, Alexa Fluor 488 (Goat polyclonal)	Thermo Fisher Scientific	A-11034	IF(1:500)
Antibody	anti-Rabbit IgG, Alexa Fluor 546 (Goat polyclonal)	Thermo Fisher Scientific	A-11035	IF(1:500)
Antibody	anti-Rabbit IgG, Alexa Fluor 647 (Goat polyclonal)	Thermo Fisher Scientific	A-21245	IF(1:500)
Antibody	anti-Rabbit IgG, Atto 594 (Goat polyclonal)	Rockland Immunochemicals Inc.	611-155-122	IF(1:500)
Antibody	anti-Rabbit IgG, Star Red (Goat polyclonal)	abberior GmbH	STRED-1002–500 UG	IF(1:500)
Cell line (Homo sapiens)	Human embryonic kidney (HEK) 293T	ATCC	CRL-3216
Chemical compound, drug	2,4,6-Triiodophenol (TIP)	Thermo Fisher Scientific	AAA1714506	4 μM
Chemical compound, drug	20 x SSC	Thermo Fisher Scientific	AM9763
Chemical compound, drug	Cycloheximide (CHX)	Sigma-Aldrich	C4859	50 nM
Chemical compound, drug	DABCO	Sigma-Aldrich	D27802
Chemical compound, drug	D-AP5	Tocris	0106	10 μM
Chemical compound, drug	Dimethyl sulfoxide (DMSO), Anhydrous	Thermo Fisher Scientific	D12345
Chemical compound, drug	Glycine	Tocris	0219	0.2 mM
Chemical compound, drug	Jasplakinolide (Jsp)	Tocris	2792
Chemical compound, drug	JF₅₄₉-HTL	Grimm et al., 2015	N/A	10 nM
Chemical compound, drug	JF₅₄₉i-HTL	Xie et al., 2017	N/A	10 nM
Chemical compound, drug	JF₆₄₆-HTL	Grimm et al., 2015	N/A	20 nM
Chemical compound, drug	Latrunculin A (LatA)	Tocris	3973	1 μM
Chemical compound, drug	Lipofectamine 3000	Thermo Fisher Scientific	L3000001
Chemical compound, drug	MNI-caged-L-glutamate (MNI)	Tocris	14-905-0	2 mM
Chemical compound, drug	Mowoil 4–88	Sigma-Aldrich	813831

Share this article

Cite this article

Chemical LTP induction reduces active transport and diffusion of GluA1-HT vesicles in the dendritic shaft.

Figure 1—source data 1

Structural LTP stimulation reduces active transport and diffusion of GluA1-HT vesicles proximal to the site of synaptic activity.

Figure 2—source data 1

Reduced motion of GluA1-HT vesicles during synaptic activity is mediated by actin polymerization in the dendritic shaft.

Figure 3—source data 1

F-actin-induced molecular crowding inhibits the motion of particles near the sites of synaptic activity.

Figure 4—source data 1

Increased GluA1-HT-SEP exocytosis triggered by synaptic activity is dependent on actin polymerization and myosin activity.

Figure 5—source data 1

GluA1-HT-SEP vesicles are drawn from a local pool in the dendritic shaft prior to exocytosis and exhibit lateral motion that is dependent on actin polymerization and myosin in the shaft during synaptic activity.

Figure 6—source data 1

Actin polymerization in the dendritic shaft proximal to the site of synaptic activity promotes GluA1 exocytosis by increasing the local pool of GluA1 vesicles and facilitating myosin-dependent transport to the dendrite periphery.

Localization and tracking parameters for the MTT program.

Author details

Victor C Wong

Contribution

For correspondence

Competing interests

Patrick R Houlihan

Contribution

Competing interests

Hui Liu

Contribution

Competing interests

Deepika Walpita

Contribution

Competing interests

Michael C DeSantis

Contribution

Competing interests

Zhe Liu

Contribution

Competing interests

Erin K O'Shea

Contribution

Competing interests

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

Categories and tags

Research organism

Further reading