Purinergic-Dependent Glial Regulation of Synaptic Plasticity of Competing Terminals and Synapse Elimination at the Neuromuscular Junction

Article

Purinergic-Dependent Glial Regulation of Synaptic Plasticity of Competing Terminals and Synapse Elimination at the Neuromuscular Junction Graphical Abstract

Authors Houssam Darabid, Alexandre St-Pierre-See, Richard Robitaille

Correspondence [email protected]

In Brief Darabid et al. show that glial cells at the neuromuscular synapse decode, in a purinergically dependent manner, the synaptic properties of nerve terminals competing for the same site. They then selectively strengthen the stronger input, which further favors its propensity to win the competition. Blockade of purinergic receptors delays synapse elimination in vivo.

Highlights d

Glial cells decode synaptic competition via P2Y1R activation and Ca2+ elevation

d

Glial cells preferentially enhance transmitter release of the stronger terminal

d

Glial cells target presynaptic A2ARs to regulate synaptic activity

d

Daily injections of a P2Y1R antagonist delays synapse elimination in vivo

Darabid et al., 2018, Cell Reports 25, 2070–2082 November 20, 2018 ª 2018 The Author(s). https://doi.org/10.1016/j.celrep.2018.10.075

Cell Reports

Article Purinergic-Dependent Glial Regulation of Synaptic Plasticity of Competing Terminals and Synapse Elimination at the Neuromuscular Junction Houssam Darabid,1,2 Alexandre St-Pierre-See,1,2 and Richard Robitaille1,2,3,* 1De ´ partement 2Groupe

de Neurosciences, Universite´ de Montre´al, PO Box 6128, Station Centre-ville, Montre´al, QC H3C 3J7, Canada de Recherche sur le Syste`me Nerveux Central, Universite´ de Montre´al, PO Box 6128, Station Centre-ville, Montre´al, QC H3C 3J7,

Canada 3Lead Contact *Correspondence: [email protected] https://doi.org/10.1016/j.celrep.2018.10.075

SUMMARY

The precise wiring of synaptic connections requires the elimination of supernumerary inputs competing for innervation of the same target cell. This competition is activity-dependent, strengthening some inputs whereas others are eliminated. Although glial cells are required for the elimination and clearance of terminals, their involvement in activity-dependent synaptic competition remains ill-defined. Here, we used the developing neuromuscular junctions of mice to show that perisynaptic glial cells, through 2Y1 purinergic receptors (P2Y1Rs), decode synaptic efficacy of competing terminals in a Ca2+-dependent manner. This glial activity induces long-lasting synaptic potentiation of strong but not weak terminals via presynaptic adenosine 2A receptors. Blockade of glial activity by intracellular Ca2+ chelation or blockade of P2Y1Rs prevents this plasticity. In addition, blockade of P2Y1Rs delays synapse elimination in vivo. Hence, P2Y1Rs drive glial cell regulation of strong synaptic inputs and influence synapse competition and elimination. INTRODUCTION The connectivity of the nervous system is shaped by a critical period of synapse competition and elimination during postnatal development. At birth, there are more synaptic contacts formed than maintained in adulthood such that a single cell initially receives an excessive number of inputs competing for its innervation (Chen and Regehr, 2000; Katz and Shatz, 1996; Lohof et al., 1996; Wyatt and Balice-Gordon, 2003). The presence of exuberant innervation and the necessity of synapse elimination have been observed in many areas of the CNS (Campbell and Shatz, 1992; Chen and Regehr, 2000; Del Rio and Feller, 2006; Hashimoto and Kano, 2013; Hooks and Chen, 2006; LaMantia and Rakic, 1990; LeVay et al., 1980; Lohof et al., 1996; Mariani and Changeux, 1981) and at the neuromuscular junction (NMJ) (Redfern, 1970; Wyatt and Balice-Gordon, 2003). This

refinement of connectivity relies on glial cells for synapse pruning in the CNS (Bialas and Stevens, 2013; Chung et al., 2013; Paolicelli et al., 2011; Schafer et al., 2012; Stevens et al., 2007; Zhan et al., 2014) and elimination of nerve terminals at the NMJ (Bishop et al., 2004; Smith et al., 2013; Song et al., 2008). At birth, each NMJ is innervated by several nerve terminals originating from distinct motor neurons (Balice-Gordon and Lichtman, 1993; Cai et al., 2013; Redfern, 1970; Tapia et al., 2012; Wyatt and Balice-Gordon, 2003). These nerve terminals undergo an activity-dependent competition, resulting in the elimination of all but one input (Bishop et al., 2004; Keller-Peck et al., 2001b; Tapia et al., 2012; Walsh and Lichtman, 2003). At dually innervated NMJs, it is accepted that the nerve terminal with a better synaptic efficacy (‘‘strong terminal’’) is more likely to be maintained, whereas the less active one (‘‘weak terminal’’) is gradually eliminated (Balice-Gordon and Lichtman, 1994; Buffelli et al., 2003; Busetto et al., 2000; Colman et al., 1997; Favero et al., 2012; Ribchester and Taxt, 1983; Ridge and Betz, 1984). However, the mechanisms responsible for the regulation of synaptic activity of terminals remain elusive. One possible candidate is the perisynaptic glial cell. Indeed, glial cells at mature synapses regulate synaptic efficacy and modulate plasticity by the release of neuroactive molecules (Castonguay and Robitaille, 2001; Henneberger et al., 2010; Panatier et al., 2006, 2011; Perea and Araque, 2007; Robitaille, 1998; Serrano et al., 2006; Todd et al., 2010). This regulation depends on their ability to detect synaptic activity (i.e., release of neurotransmitters), causing intracellular Ca2+ elevations and the decoding of neuronal properties (Todd et al., 2010). Interestingly, we recently showed that, during NMJ development, persiynaptic schwann cells (PSCs) decode the synaptic activity of competing inputs, via purinergic type 2Y receptors, based on differential synaptically induced intracellular Ca2+ elevations (Darabid et al., 2013). Furthermore, in purinergic type 2Y1 receptor (P2Y1R) knockout mice, synaptically induced Ca2+ elevations in PSCs are obliterated (Heredia et al., 2018). PSCs’ ability to decode synaptic activity could endow them with the capacity to differentially regulate the synaptic strength of competing terminals and, thus, influence the outcome of synaptic competition and elimination. Here we tested this possibility by combining simultaneous Ca2+ imaging of PSCs and synaptic recordings of dually

2070 Cell Reports 25, 2070–2082, November 20, 2018 ª 2018 The Author(s). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

innervated NMJs from post-natal day 7 (P7)–P8 pups. We show that stronger terminals displayed long-lasting potentiation of synaptic activity in response to high-frequency stimulation (HFS), whereas it was smaller and transient in the weaker terminal. This plasticity was mediated by presynaptic adenosine type 2A receptors (A2ARs). Also, it was obliterated by the blockade of PSC Ca2+ activity using Ca2+-chelating molecules or by blocking PSC P2Y1Rs, whereas PSC-specific Ca2+ elevations generated the same pattern of plasticity. Importantly, in vivo blockade of P2Y1Rs delayed synapse elimination. Hence, we present a glial cell purinergic mechanism that preferentially potentiates stronger nerve terminals and influences synapse competition and elimination. RESULTS Because of the Ca2+-dependence of PSCs functions, we monitored and perturbed PSC intracellular Ca2+ signaling to study their role in the regulation of synaptic activity in dually innervated NMJs from P7–P8 mice and their effect on activity-dependent synaptic competition and elimination. Single-Cell Electroporation of PSCs Does Not Alter Their Activity and Synaptic Transmission We used single-cell electroporation to specifically load PSCs at dually innervated NMJs with different charged molecules, including the calcium indicator Fluo-4 and the fluorescent dye Alexa Fluor 594, to monitor and alter PSCs’ Ca2+ activity. This approach allowed us to rapidly and reliably load multiple PSCs in sequence (Figure S1; Video S1). Importantly, PSC electroporation did not affect the amplitude of endplate potentials (EPPs) generated by each competing terminal (see Figure S2 and below for synaptic strength measurements), which remained stable for over 50 min (strong inputs: 101.8% ± 3.4% before versus 100.9% ± 7.6% after 4 min, 95.9% ± 1.8% after 20 min, 95.7% ± 4.9% after 40 min, N = 5, one-way ANOVA, repeated measures, Tukey’s multiple comparisons posttest, p > 0.05; weak inputs: 102.9% ± 3.5% before versus 100.0% ± 6.4% after 4 min, 94.1% ± 5.1% after 20 min, 97.2% ± 7.6% after 40 min, N = 5, one-way ANOVA, repeated measures, Tukey’s multiple comparisons post-test, p > 0.05; Figures S1D–S1F). Finally, PSCs’ responsiveness to local transmitter applications was also unaffected (see below; Figure 1; Figure S1). Hence, it is a reliable method to specifically load PSCs at developing NMJs that does not alter the synaptic activity of competing terminals. PSCs Decode Synaptic Competition PSCs must detect and differentiate the synaptic activity of each competing nerve terminal to provide an appropriate feedback (Todd et al., 2010). This was shown in our previous study (Darabid et al., 2013), and we proceeded to confirm it based on the use of single-cell electroporation for loading PSCs. The identity of the stronger and weaker inputs was established based on their respective synaptic strength determined by the quantal content (m) using the failure method (Del Castillo and Katz, 1954) and facilitation (F) using a paired-pulse F protocol (Figure S2; Darabid et al., 2013; Kopp et al., 2000). Of the two competing inputs at a

dually innervated NMJ, the strong terminal is characterized by the largest m and the least paired-pulse F (Darabid et al., 2013; Kopp et al., 2000). We loaded the calcium indicator Fluo-4 and the fluorescent dye Alexa Fluor 594 into all visible PSCs at dually-innervated NMJs (Darabid et al., 2013; Hirata et al., 1997; Kopp et al., 2000). Using confocal Ca2+ imaging, we studied PSCs’ responsiveness to endogenous synaptic activity from each competing input induced by HFS of a specific ventral root (50-Hz continuous stimulation for 30 s). This stimulation pattern was designed based on reports describing the in vivo tonic firing of motor neurons innervating the soleus muscle (Eken et al., 2008; Gorassini et al., 2000). Interestingly, at dually innervated NMJs, HFS of the stronger nerve terminal (i.e., the terminal with the larger m value) always induced larger PSC Ca2+ responses than its weaker competitor (Figures 1A, 1B, 1D, and 1E; strong: 39.53% ± 13.65% fluorescence intensity ratio [DF/F0] versus weak: 21.43% ± 10.95% DF/F0; N = 8, paired t test, p < 0.05). This shows that a PSC detects neurotransmitter release from each competing nerve terminal and decodes its synaptic strength. These results using single-cell electroporation are similar to those using bulk loading of the Ca2+ indicator in PSCs (Darabid et al., 2013). Moreover, PSCs also responded to the local application of ATP (5 mM; Figure 1C; 131.50% ± 48.19% DF/F0; N = 4), which is the main neurotransmitter activating PSCs at developing NMJs (Darabid et al., 2013; Heredia et al., 2018) and is commonly used to assess PSC excitability and health (Arbour et al., 2015; Darabid et al., 2013; Rochon et al., 2001). Altogether, these results confirm that PSCs at a dually innervated NMJ decode synaptic competition and that PSCs can be targeted specifically by single-cell electroporation without undesired effects. Only Strong Competing Nerve Terminals Display LongLasting Potentiation of Synaptic Activity Because activity is a major determinant for synapse competition and elimination, we hypothesized that competing terminals may display different levels of synaptic plasticity that could reflect their respective synaptic strength. Because stronger terminals are favored to win the competition at dually innervated NMJs (Buffelli et al., 2003; Colman et al., 1997; Kopp et al., 2000), we predict that they are more prone to produce potentiation than weaker ones. HFS of strong inputs, which induced large PSC Ca2+ elevations, elicited a short-term potentiation (calculated 4 min after HFS; EPP amplitude of 170.7% ± 16.7% versus 99.1% ± 1.8% at baseline; N = 8, one-way ANOVA, repeated measures, Tukey’s multiple comparisons post-test, p < 0.05), followed by a long-lasting potentiation (calculated 16 min after HFS; Figures 1F and 1H) that persisted for at least the duration of the recordings (EPP amplitude of 145.4% ± 19.1% calculated 16 min postHFS; N = 8, significantly different from baseline, 99.1% ± 1.8%, one-way ANOVA, repeated measures, Tukey’s multiple comparisons post-test, p < 0.05). However, the HFS of weak inputs, which induced small PSC Ca2+ elevations, only elicited a shortterm increase in EPP amplitude (EPP amplitude of 120.9% ± 4.9% 4 min post-HFS versus 101.2% ± 1.4% during the baseline; N = 7, one-way ANOVA, repeated measures, Tukey’s

Cell Reports 25, 2070–2082, November 20, 2018 2071

Figure 1. Competing Nerve Terminals Induce Different Ca2+ Responses in PSCs and Show Distinct Synaptic Plasticity to High-Frequency Stimulation (A–C) Average of Ca2+ responses ± SEM (dotted lines) induced in PSCs by independent stimulation of the strong (green in A, N = 8) and weak (blue in B, N = 8) competing inputs and local application of ATP (5 mM, gray in C N = 4). A false-color confocal image of the electroporated PSC (Alexa Fluor 594 in red; scale bar, 10 mm) as well as images of changes in the fluorescence of Fluo-4, illustrating Ca2+ levels before (1), at the peak of the response (2), and after stimulation (3) are presented for each condition. (D) Plot of the amplitude of PSC Ca2+ responses induced by strong (green dots) and weak (blue dots) terminals. Each connected pair of strong and weak terminals, which induced Ca2+ responses in the same PSC (1 PSC/NMJ), are in competition at the same NMJ. Note that strong inputs always induced larger PSC Ca2+ responses than weak inputs. (E) Histograms showing the mean amplitude of the PSC Ca2+ responses ± SEM induced by strong (green) and weak (blue) inputs. (F and G) Normalized EPP amplitude ± SEM over time, showing that stimulation (black arrow at time 0) of stronger inputs resulted in a robust long-lasting potentiation of neurotransmission (green in F, N = 8), whereas stimulation of weaker inputs resulted in a transient potentiation (blue in G, N = 7). Each point represents a 2-min mean of 24 EPPs (recorded at 5-s intervals). The insets show examples of EPPs recorded before (1, baseline), shortly after the HFS (2, at 4 min), and 16 min after the HFS. (H and I) Histograms showing the mean amplitude of EPPs ± SEM induced by strong (green in H) and weak (blue in I) inputs during the baseline, shortly after the HFS (at 4 min) and 16 min after the HFS. *p < 0.05; **p % 0.01. See also Figures S1, S2, and S5.

multiple comparisons post-test, p < 0.05), which returned to baseline after a few minutes (Figures 1G and 1I; EPP amplitude of 99.3% ± 3.3% 16 min post-HFS; N = 7, was not significantly different from baseline 101.2% ± 1.4%, one-way ANOVA, repeated measures, Tukey’s multiple comparisons post-test, p > 0.05). Hence, strong and weak terminals display distinct levels of synaptic plasticity following HFS, where the long-lasting potentiation of stronger terminals represents an enhancement of synaptic strength. This hints to the presence of mechanisms that selectively increase the activity of the stronger but not the weaker input. PSC Ca2+ Activity Is Necessary and Sufficient to Elicit Synaptic Plasticity of Competing Nerve Terminals Because PSC Ca2+-activity regulates synaptic plasticity at mature NMJs (Castonguay and Robitaille, 2001; Todd et al., 2010), we hypothesized that it is required for the synaptic poten-

2072 Cell Reports 25, 2070–2082, November 20, 2018

tiation of competing terminals. To test this hypothesis, we blocked PSC Ca2+ activity using Diazo2 electroporated into PSCs. Diazo2 is a caged BAPTA (1,2-Bis(2-aminophenoxy) ethane-N,N,N0 ,N0 -tetraacetic acid) molecule with a very low affinity for Ca2+ at rest (Kamiya and Zucker, 1994). When photoactivated (exposure to 405-nm laser light), its affinity to Ca2+ increases greatly, providing rapid buffering of Ca2+ and interfering with PSC activity (Kamiya and Zucker, 1994; Todd et al., 2010). Following blue light exposure in the absence of Diazo2, or when Diazo2 is present but not photoactivated, PSCs responded to endogenous transmitter release from nerve terminals and to local applications of ATP (Figure S3). However, similar to previous work at the mature NMJ (Todd et al., 2010), photoactivation of Diazo2 prevented PSC Ca2+ elevations induced by local applications of ATP (5 mM), and this chelation was specific to the photoactivated cells loaded with Diazo2 (Figure S3B). It is noteworthy that, although Ca2+ chelation was efficient, the Diazo2-dependent blockade could be reversed when

Figure 2. Blockade of PSC Ca2+ Activity Alters the Synaptic Plasticity of Competing Terminals (A–C) Photolysis of caged BAPTA molecules (Diazo2) electroporated specifically in PSCs prevents PSC Ca2+ responses. Shown is the average Ca2+ activity ± SEM (dotted lines) induced in PSCs by independent stimulation of the strong (green in A, N = 5) and weak competing inputs (blue in B, N = 5) and local application of ATP (5 mM, gray in C, N = 5) following photo-activation of Diazo2. Note that no PSC Ca2+ responses were induced. A false-color confocal image of the electroporated PSC (Alexa Fluor 594 in red; scale bar, 10 mm) as well as images of changes in fluorescence of electroporated Fluo-4, illustrating Ca2+ levels before (1), at the peak of the response (2), and after stimulation (3), are presented for each condition. (D) Plot of the amplitude of PSC Ca2+ responses induced by strong (green dots) and weak (blue dots) terminals. Each connected pair of strong and weak terminals is in competition at the same NMJ, whose Ca2+ responses are recorded in the same PSC (1 PSC/NMJ). (E) Histograms showing the mean amplitude of the PSC Ca2+ responses ± SEM induced by strong (green) and weak (blue) inputs. (F and G) Normalized EPP amplitude ± SEM over time, showing that stimulation (gray arrow at time 0) of strong (green in F, N = 4) or weak (blue in G, N = 6) inputs no longer resulted in changes of synaptic activity following blockade of PSC Ca2+ responses. The insets show examples of EPPs recorded before (1, baseline), shortly after the HFS (2, at 4 min), and 16 min after the HFS. (H and I) Histograms showing the mean amplitude of EPPs ± SEM induced by strong (green in H) and weak (blue in I) inputs during the baseline, shortly after the HFS (at 4 min) and 16 min after the HFS. See also Figure S3.

PSC intracellular photoactivated Diazo2 was saturated by inducing successive Ca2+ rises (local applications of ATP; Figure S3C). This also reveals that the activation of Diazo2 did not completely shut down PSC activity and did not alter their viability. Diazo2 efficiently blocked PSC Ca2+ activity induced by exogenous and endogenous transmitters (Figure 2). Indeed, neither HFS stimulation of strong inputs nor of weak ones induced Ca2+ elevations in PSCs following photoactivation of Diazo2 (strong: 4.02% ± 0.98% DF/F0; weak: 3.53% ± 1.07% DF/F0; N = 5). Moreover, a single local application of ATP (5 mM) did not induce any Ca2+ elevation in PSCs (Figure 2C; control [CTRL]: 131.50% ± 48.19% DF/F0, N = 4 versus photoactivated Diazo2: 3.74% ± 0.52% DF/F0, N = 5; unpaired t test, p = 0.03). Importantly, blockade of Ca2+ elevations in PSCs by photoactivated Diazo2 (Figure S3A) prevented the HFS-induced differential plasticity of competing terminals. Indeed, no persistent potentiation of the strong terminal was observed following blockade of PSC Ca2+ activity (Figures 2F and 2H; EPP ampli-

tude of 98.0% ± 2.0% at baseline versus 112.2% ± 7.9% 4 min post-HFS versus 91.7% ± 12.1% 16 min post-HFS; N = 4, all non-significant using one-way ANOVA, repeated measures, Tukey’s multiple comparisons post-test, p > 0.05). Interestingly, the synaptic plasticity of weak terminals was unaltered by blockade of PSC Ca2+ responses (Figures 2G and 2H; EPP amplitude of 101.3% ± 3.1% during the baseline versus 110.4% ± 16.5% 4 min post-HFS versus 83.8% ± 10.1% 16 min post-HFS; N = 6, all non-significant using one-way ANOVA, repeated measures, Tukey’s multiple comparison post-test, p > 0.05). This shows that weaker nerve terminals receive little modulation from PSCs even though they can elicit glial Ca2+ elevations (Figures 1 and 2). In addition, these results reveal that PSC Ca2+ activity is required to generate synaptic plasticity of the stronger input. Moreover, the complete blockade of this plasticity by PSC Ca2+ chelation strongly argues in favor of PSC-dependent mechanisms. Indeed, if the observed effect was only partially due to PSC Ca2+ activity, then chelating calcium would not have completely blocked the plasticity of strong competing terminals.

Cell Reports 25, 2070–2082, November 20, 2018 2073

Figure 3. Direct Larger, but Not Smaller, PSC Ca2+ Elevation Induces Synaptic Plasticity (A and B) Mean amplitude of larger (red in A, N = 4) and smaller (red in B, N = 4) PSC Ca2+ responses induced by the photoactivation of NP-EGTA electroporated specifically in PSCs (photoactivated by a single 405-nm laser pulse of 500-ms duration). The gray dotted lines represent ± SEM. A falsecolor confocal image of the electroporated PSC (Alexa Fluor 594 in red, scale bar 10 mm) as well as images of changes in fluorescence of electroporated Fluo-4, illustrating Ca2+ levels before, at the peak of the response, and after photoactivation of NP-EGTA, are presented. (C and D) Normalized EPP amplitude ± SEM over time, showing that larger (C) but not a smaller (D) PSC Ca2+ responses, induced by photoactivation of NP-EGTA, resulted in the plasticity of strong (green) but not weak (blue) inputs. The insets show examples of EPPs recorded before (1, baseline) and 22 min after NP-EGTA activation (2). (E and F) Histograms showing the mean amplitude of EPPs ± SEM of strong (E, N = 4) and weak (F, N = 4) inputs induced by larger PSC Ca2+ responses during the baseline and 22 min after photoactivation of NP-EGTA. (G and H) The same as (E) (G, strong, N = 4) and (F) (H, weak, N = 4) but for smaller NP-EGTA-induced PSC Ca2+ responses. *p < 0.05.

We next directly activated PSCs using a caged Ca2+ molecule (nitrophenyl [NP]-EGTA) to test whether PSC Ca2+ elevations are sufficient to potentiate synaptic activity. As shown in Figure 3, inducing a larger Ca2+ elevation in PSCs (68.87% ± 22.91% DF/F0, N = 4) by NP-EGTA photolysis was sufficient to induce a long-lasting potentiation of the stronger input (EPP amplitude of 101.0% ± 0.9% during the baseline versus 129.6% ± 8.1% at 22 min following NP-EGTA photolysis; N = 4, paired t test, p < 0.05). However, this did not affect the synaptic activity of the weak input (Figures 3A and 3C; EPP amplitude of 101.3% ± 1.7% during the baseline versus 106.8% ± 7.1% following NP-EGTA photolysis; N = 4, non-significant, paired t test, p > 0.05). However, inducing a smaller PSC Ca2+ elevation with NP-EGTA (27.92% ± 7.91% DF/F0) had no significant effect on the synaptic activity of either competing terminal (Figures 3B and 3D; EPP amplitude of strong inputs 99.5% ± 3.0% during the baseline versus 101.1% ± 2.2% following NP-EGTA photolysis; N = 4, paired t test, p > 0.05; EPP amplitude of weak inputs 101.3% ± 0.1% during the baseline versus 95.1% ± 5.3% following NP-EGTA photolysis; N = 4, paired t test, p > 0.05). These results indicate that larger photoactivation-induced Ca2+ elevations in PSCs were sufficient to induce long-lasting synaptic potentiation of stronger inputs, leaving weaker inputs unaffected. This is consistent with a Ca2+-dependent threshold

2074 Cell Reports 25, 2070–2082, November 20, 2018

mechanism whereby a certain level of Ca2+ must be reached to observe the potentiation. Altogether, these results indicate that PSC Ca2+ activity is both necessary and sufficient to differentially regulate the expression of synaptic plasticity of stronger terminals over weaker ones. Synaptic Potentiation Is Mediated by PSC P2Y1Rs and Presynaptic A2ARs We next investigated the mechanisms by which PSCs differentially regulate competing inputs. During synaptic competition, PSC Ca2+ activity is mediated by P2YRs activated by ATP (Darabid et al., 2013; Heredia et al., 2018). Interestingly, although their presence on other synaptic elements cannot be ruled out, P2Y1Rs are located on PSCs during synaptic competition, clustered near presynaptic release sites (Darabid et al., 2013). Hence, we tested the contribution of P2Y1Rs on PSCs using MRS2179 (20 mM), a specific P2Y1R antagonist (Boyer et al., 2002). Interestingly, no Ca2+ elevations were elicited by local application of ATP in the presence of MRS2179 (Figures 4A and 4B; 10.95% ± 1.90% DF/F0, n = 20 PSCs). This blockade was reversible because Ca2+ elevations were induced after a 20-min washout of the drug (300.20% ± 12.00% DF/F0, n = 20 PSCs; paired t test, p < 0.0001). This is consistent with recent work showing that purine-induced PSC Ca2+ responses are mediated exclusively by P2Y1Rs (Heredia et al., 2018) and further confirms that ATP-dependent activation of PSCs is mediated by P2Y1Rs.

Figure 4. The PSC Ca2+ Activity and Synaptic Plasticity of Competing Terminals Depend on P2Y1Rs (A) Examples of PSC Ca2+ responses to local application of ATP (5 mM) before (left), during (center), and after washout (right) of the P2Y1Rs antagonist (MRS2179, 20 mM). Black and gray traces represent 2 PSCs at a dually innervated NMJ. (B) Histograms showing the mean amplitude of the PSC Ca2+ responses ± SEM induced by local application of ATP in the presence (white) or absence (black) of MRS2179 (n = 20). (C and D) Normalized EPP amplitude ± SEM over time, showing that stimulation (black arrow at time 0) of strong (green in C, N = 6) or weak (blue in D, N = 6) inputs no longer resulted in changes of synaptic activity in the presence of MRS2179. The insets show examples of EPPs recorded before (1, baseline), shortly after the HFS (2, at 4 min), and after the HFS (3, at 16 min). (E and F) Histograms showing the mean amplitude of EPPs ± SEM induced by strong (green in E) and weak (blue in F) inputs during the baseline, shortly after the HFS (at 4 min), and 16 min after the HFS. *p < 0.05; **p % 0.01; ***p % 0.001. See also Figure S5.

Because synaptic plasticity depends on PSC Ca2+ activity, and PSCs activation is mediated by P2Y1Rs, blocking those receptors should alter the endogenous plasticity of terminals. Indeed, bath application of MRS2179 (20 mM) prevented the long-lasting potentiation of strong inputs following HFS (Figures 4C and 4E; EPP amplitude of 97.5% ± 3.0% during the baseline versus 86.2% ± 5.4% 16 min post-HFS; N = 6, one-way ANOVA, repeated measures, Tukey’s multiple comparisons post-test, p > 0.05). In addition, the use of MRS2179 unravelled a depression of EPP amplitude observed 16 min post-HFS, particularly for the weak inputs (Figures 4D and 4F; EPP amplitude of 101.8% ± 0.8% during the baseline versus 109.0% ± 2.9% 4 min postHFS versus 81.5% ± 8.9% 16 min post-HFS; N = 6, one-way ANOVA, repeated measures, Tukey’s multiple comparisons post-test, p < 0.05). As a whole, these data show that P2Y1Rs mediate the ATP-dependent activation of PSCs and that the potentiation of synaptic activity of stronger competing terminals depends on these receptors. Although the contribution of nonglial P2Y1Rs cannot be completely ruled out using pharmacological approaches, the similarity of these results to the necessity and sufficiency of PSC Ca2+ activity for potentiating stronger competing terminals argues in favor of a PSC-autonomous effect.

We next investigated which presynaptic receptors were responsible for the potentiation of synaptic transmission. At mature NMJs, PSC-dependent potentiation of synaptic activity is mediated by presynaptic A2ARs (Todd et al., 2010). These receptors have been located on nerve terminals of adults and new born mice (Garcia et al., 2013; Toma`s et al., 2014). Although there is evidence that A2ARs are also expressed by glial cells (Alloisio et al., 2004; Hettinger et al., 2001; Matos et al., 2015; Toma`s et al., 2014), we show that, unlike ATP, local applications of adenosine (10 mM) failed to induce any Ca2+ elevation in PSCs at all tested NMJs (Figures 5A and 5B; adenosine: 6.5% ± 1.34% DF/F0 versus ATP: 101.20% ± 18.50% DF/F0; 8 PSCs, 5 NMJs, 5 muscles, paired t test, p = 0.0017). This suggests that direct PSCs activation by adenosine is unlikely and that their action must be downstream of PSCs activation. However, bath application of a specific A2ARs antagonist, SCH58261 (100 nM), prevented potentiation of the stronger nerve terminal induced by HFS (Figures 5C and 5E; EPP amplitude of 100.6% ± 4.85% during baseline versus 98.38% ± 3.47% 4 min post-HFS versus 90.83% ± 14.39% 16 min postHFS; N = 4, one-way analysis of variance, Tukey’s multiple comparisons post-test, p > 0.05). Also, EPP amplitude following HFS

Cell Reports 25, 2070–2082, November 20, 2018 2075

Figure 5. The Synaptic Plasticity of Competing Terminals Depends on A2ARs (A) Example of a PSC Ca2+ response to local application of adenosine (10 mM, left) and ATP (5 mM, right). Unlike ATP, adenosine failed to induce any Ca2+ elevation in PSCs. (B) Histograms showing the mean amplitude of the PSC Ca2+ responses ± SEM induced by local application of adenosine (white) and ATP (black) (n = 8, N = 5) . (C and D) Normalized EPP amplitude ± SEM over time, showing that stimulation (gray arrow at time 0) of strong (green in C, N = 4) or weak (blue in D, N = 4) inputs no longer resulted in changes of synaptic activity in the presence of the A2AR antagonist SCH58261 (100 nM). The insets show examples of EPPs recorded before (1, baseline), shortly after the HFS (2, at 4 min), and after the HFS (3, at 16 min). (E and F) Histograms showing the mean amplitude of EPPs ± SEM induced by strong (green in E) and weak (blue in F) inputs during the baseline, shortly after the HFS (at 4 min) and 16 min after the HFS. **p % 0.01. See also Figures S4 and S5.

of the weaker input was not different from baseline during bath application of SCH58261 (Figures 5D and 5F; EPP amplitude of 100.4% ± 1.81% during baseline versus 109.5% ± 7.59% 4 min post-HFS versus 105.9% ± 2.92% 16 min post-HFS; N = 4, one-way analysis of variance, Tukey’s multiple comparisons post-test, p > 0.05). These data suggest that A2AR activation is downstream of PSCs activation and is necessary for the endogenous potentiation of stronger competing terminals. Interestingly, bath application of the specific A2AR agonist CGS21680 (7 nM) (Correia-de-Sa´ et al., 1996; Oliveira et al., 2004; Todd et al., 2010) potentiated neurotransmission from both competing terminals (Figure S4; EPP amplitude of strong inputs 103.9% ± 3.82% during baseline versus 142.7% ± 2.92% 16 min post-HFS; N = 4, paired t test, p = 0.0068; EPP amplitude of weak inputs 98.3% ± 1.70% during baseline versus 136.7% ± 3.23 16 min post-HFS; N = 4, paired t test, p = 0.0037). These data imply that both terminals can be potentiated regardless of their synaptic strength. Moreover, this indicates that the mechanisms that enable the specific targeting of input potentiation are upstream of a presynaptic terminal’s activation, making PSCs prime candidates. Altogether, these results suggest that the differential plasticity of competing nerve terminals depends on PSC P2Y1Rs and on the activation of presynaptic A2ARs.

2076 Cell Reports 25, 2070–2082, November 20, 2018

In Vivo Blockade of P2Y1Rs Delays Synapse Elimination Our results indicate that PSCs can differentially modulate synaptic activity of competing nerve terminals. However, it is unknown whether such a PSC-dependent mechanism would influence synaptic competition and elimination. To this end, we targeted P2Y1Rs in vivo because of their role in PSCs activation and differential synaptic potentiation (Figures 2 and 4). We performed daily subcutaneous injections (from P4 to P14; Figure 6A) of a specific P2Y1Rs antagonist, MRS2179 (40 mM), in the hindlimb area next to the soleus muscle. The contralateral limb was injected with physiological saline. Animals were then sacrificed at either P8, P10, P12 or P14. This time window allows analysis of synapse elimination progression (Kopp et al., 2000; Personius and Balice-Gordon, 2001; Personius et al., 2007), including the period of differential synaptic plasticity of terminals (P7–P8). Unlike constitutive knockout (KO) of P2Y1Rs (Heredia et al., 2018), the ages we targeted allow fetal and early post-natal maturation of NMJs (
Figure 6. In Vivo Blockade of P2Y1Rs Delays Synapse Elimination (A) Schematic representation of subcutaneous MRS2179 injection. Daily subcutaneous injections of 40 mM MRS2179 were performed next to the soleus muscle from P4 to P14. Mice were scarified at P8, P10, P12, and P14. (B–E) Confocal images presenting examples of saline-injected (first column, B1–E1) and MRS2179-injected soleus (second column, B2– E2), labeled to observe presynaptic nerve terminals (stained with antibodies against NF-M and SV2, green) and postsynaptic endplates (nAChRs stained with a-bungarotoxin, red). Note that more polyinnervated NMJs (asterisks) were present in MRS2179-injected mice at all ages. The state of polyinnervation was defined by the number of independent nerve terminals that innervate the same endplate. B3–E3 (third column) show a higher magnification of polyinnervated NMJs, highlighted by a rectangle in B1 and B2 to E1 and E2. Independent inputs (green) innervating the same endplate area (red) are indicated by white arrows. White squares and dots represent the enlarged region from saline-injected mice, whereas red square and dots represent the enlarged region from MRS2179-injected mice. Note that most P14 saline-injected mice (E1 and E3, blue square and dot) were mono-innervated by a single input, marked by a white arrowhead. (F) Diagram showing the time course of synapse elimination in saline-injected (black) compared with MRS2179-injected (red) mice at P8 (Nsaline = 8, NMRS = 8), P10 (Nsaline = 8, NMRS = 8), P12 (Nsaline = 7, NMRS = 7), and P14 (Nsaline = 7, NMRS = 6). Note that MRS2179-injected mice had a delay in synapse elimination, highlighted by the presence of more polyinnervated NMJs at P12 and P14. (G) Histograms highlighting the difference in the percentage of polyinnervation between saline-injected (black) and MRS2179-injected (red) mice at P12 and P14. *p < 0.05; ***p % 0.001. Scale bars, 10 mm.

[NF-M] that contacted the endplate area of a single muscle fiber (labeled postsynaptic nicotinic cholinergic receptors [nAChRs] with a-bungarotoxin; Figures 6B–6E). In contralateral muscles injected with saline, polyinnervation decreased drastically from P8, with more than 50% of polyinnervated NMJs, to P14, where mono-innervation prevailed (Figures 6B–6G; 57.6% ± 4.3% of polyinnervated NMJs at P8, 8 muscles, 8 mice; 26.3% ± 4.9% at P10, 8 muscles,

8 mice; 11.5% ± 2.2% at P12, 7 muscles, 7 mice; 1.8% ± 0.9% at P14, 7 muscles, 7 mice). These data are in accordance with numbers reported at the same muscle and ages studied (Kopp et al., 2000; Personius and Balice-Gordon, 2001; Personius et al., 2007). No differences were observed between non-injected and saline-injected muscles (data not shown). Unlike saline injections, in vivo blockade of P2Y1Rs altered polyinnervation at different ages (Figures 6B–6G). First, no changes in polyinnervation were observed at P8 or P10 (P8 MRS2179-injected: 60.7% ± 2.7%, 8 muscles, 8 mice versus saline-injected: 57.6% ± 4.3%, 8 muscles, 8 mice; unpaired t test, p > 0.05; P10 MRS2179 injected: 38.6% ± 5.4%, 8 muscles, 8 mice versus saline-injected: 26.3% ± 4.9%, 8 muscles, 8 mice; unpaired t test, p > 0.05). However, at P12, we

Cell Reports 25, 2070–2082, November 20, 2018 2077

Figure 7. Morphological Signs of Ongoing Synapse Elimination at NMJs from P14 MRS2179-Injected Mice (A and B) Confocal images of NMJs from P8 (A1, N = 8) and P14 (A2, N = 7) saline-injected and P8 (B1, N = 8) and P14 (B2, N = 6) MRS2179-injected soleus muscles. Presynaptic nerve terminals were labeled with antibodies against NF-M and SV2 (green), and postsynaptic endplates were labeled with a-bungarotoxin (red). The insets show a higher magnification of NMJs, highlighted by a rectangle in (A1) and (B1) and (B2). White squares and dots represent the enlarged region from saline-injected mice, whereas red square and dots represent the enlarged region from MRS2179-injected mice. The presence of retraction bulbs (enlargement of a disconnected nerve terminal in green) at a distance from the endplate area (red) is indicated by white arrows. Note that retraction bulbs (white arrows) were still observed in NMJs of P14 MRS2179-injected mice (B2). Scale bars, 10 mm.

dependent activation promotes the process of elimination and contributes to proper connectivity at the NMJ. DISCUSSION

observed a striking 3-fold greater level of polyinnervated NMJs in mice injected with MRS2179 compared with salineinjected ones (MRS2179-injected: 31.0% ± 3.6%, 7 muscles, 7 mice versus saline-injected: 11.5% ± 2.2%, 7 muscles, 7 mice; Mann-Whitney test, p = 0.0006). This increase in polyinnervation was also observed at P14, an age when synapse elimination is mostly complete (Kopp et al., 2000; Personius and Balice-Gordon, 2001; Personius et al., 2007; Figures 6F and 6G; 7.1% ± 1.9% of polyinnervated NMJs/ muscle at P14 MRS2179-injected mice, 7 muscles, 7 mice versus 1.8% ± 0.9% in saline-injected ones, 6 muscles, 6 mice; Mann-Whitney test, p = 0.0365). This suggests that P2Y1Rs blockade delays the process of synapse elimination so that the level of polyinnervation from MRS2179-injected mice at P12 and P14 is, respectively, like the one at P10 and P12 from control animals. Finally, we assessed the presence of swollen axonal tips called ‘‘retraction bulbs,’’ which are indicative of nerve terminals recently retracted from the endplate area. These are signs of ongoing postnatal synapse elimination (Balice-Gordon and Lichtman, 1993; Bishop et al., 2004). In saline-injected animals, retraction bulbs were observed at P8 but not at P14 (Figure 7A), the latter being the end stage of synapse elimination in the soleus muscle (Kopp et al., 2000; Personius and Balice-Gordon, 2001; Personius et al., 2007). However, retraction bulbs were observed at all ages, including P14, in MRS2179-injected mice (Figure 7B). This strongly suggests that synapse elimination was still ongoing, further confirming a delay in synapse elimination. Altogether, these data suggest that P2Y1Rs-

2078 Cell Reports 25, 2070–2082, November 20, 2018

In this study, we showed that PSCs decode the synaptic efficacy of competing terminals via P2Y1R activation and preferentially potentiate the stronger input, leaving the weaker one unaltered. This differential plasticity depends on glial Ca2+ activity. Preventing PSC activation alters synaptic plasticity in situ, and blocking P2Y1Rs results in delayed synapse elimination in vivo. PSCs actively reinforce the strong input that is favored to win the competition (Buffelli et al., 2003; Colman et al., 1997; Kopp et al., 2000) and, as suggested by the in vivo P2Y1R blockade, then influence synapse elimination. Hence, the activity of nerve terminals can be linked to the glia-mediated regulation of synapse competition and elimination. Synaptic Plasticity of Competing Terminals Strong terminals showed persistent potentiation of neurotransmission in response to HFS, increasing synaptic strength, and, hence, the disparity of synaptic efficacy between competing terminals. This would bias the competition toward stronger inputs because disparity in synaptic strength is a predictor of the outcome of synapse elimination (Buffelli et al., 2003; Colman et al., 1997; Kopp et al., 2000). Importantly, the HFS pattern used is similar to the endogenous activity recorded from newborn rodents during the tonic firing of motor units innervating the soleus muscle, suggesting that such preferential potentiation of strong terminals may occur in vivo (Eken et al., 2008; Gorassini et al., 2000). The propensity of strong inputs to generate larger potentiation may be responsible, at least in part, for their reinforcement and maintenance and is consistent with evidence that more active and efficient nerve terminals are favored to win the competition (Balice-Gordon and Lichtman,

1994; Buffelli et al., 2003; Busetto et al., 2000; Favero et al., 2012; Je et al., 2012; Ribchester and Taxt, 1983; Ridge and Betz, 1984; Schafer et al., 2012; Stellwagen and Shatz, 2002; Stevens et al., 2007). For instance, at the Drosophila NMJ, an acute increase of synaptic activity stimulates the formation of new synaptic boutons and synapse expansion (Ataman et al., 2008; Fuentes-Medel et al., 2009). Moreover, long-term potentiation in the CNS has been associated with synapse enlargement and stabilization (Matsuzaki et al., 2004; Na¨gerl et al., 2004). Thus, given the direct link between the activity, the plasticity, and the structure of synapses, a differential plasticity of terminals may promote synapse growth and stabilization of stronger but not weaker inputs. PSCs Govern the Synaptic Plasticity of Competing Terminals The differential potentiation of nerve terminals by PSCs depends on their ability to detect neurotransmission. Similar to previous studies (Darabid et al., 2013; Heredia et al., 2018), we confirmed that this detection depends on the activation of P2Y1Rs. In the CNS, P2Y1Rs have been observed in the processes of astrocytes surrounding excitatory synapses of the molecular layer of the dentate gyrus (Jourdain et al., 2007), where they mediate Ca2+-elevations and allow astrocytes to control synaptic strength and certain forms of plasticity (Santello et al., 2011). Moreover, astrocytes can release ATP in an inositol 3-phosphate receptor 2 (IP3R2)-dependent manner to regulate synapse elimination (Yang et al., 2016). Finally, synapse pruning in the CNS is driven by synaptic activity so that stronger inputs are more likely to be maintained (Schafer et al., 2012; Stellwagen and Shatz, 2002). PSC potentiation of strong competing inputs is consistent with glial cell regulation of synaptic plasticity (Henneberger et al., 2010; Navarrete and Araque, 2010; Navarrete et al., 2012; Santello et al., 2011; Serrano et al., 2006; Todd et al., 2010). However, even when the HFS of weaker terminals induced Ca2+ elevations in PSCs, there was little feedback regarding their synaptic activity, and no persistent potentiation of transmission was observed (Figures 1 and 2). This suggests that Ca2+ in PSCs must reach a threshold to trigger synaptic potentiation. Moreover, stronger terminals seem to be favored as weaker inputs never showed a persistent increase in neurotransmitter release even when inducing a larger PSC Ca2+ response with NP-EGTA (Figures 3C and 3F). A small contribution of P2Y1Rs to weaker terminals cannot be ruled out because their blockade resulted in a slight synaptic depression (Figure 4). However, it remains unclear whether this result suggests an active strengthening of basal synaptic transmission of weaker inputs by PSCs. The involvement of presynaptic A2ARs in the potentiation of stronger inputs is consistent with glial regulation at mature NMJs (Todd et al., 2010), where a continuous stimulation, similar to the one used in this study, potentiated neurotransmission via presynaptic A2AR activation by adenosine (Oliveira et al., 2004; Todd et al., 2010). Adenosine is often generated by the degradation of ATP (Dunwiddie et al., 1997; Rebola et al., 2008), a major gliotransmitter at mature NMJs and in the CNS (Panatier et al., 2011; Pascual et al., 2005; Serrano et al., 2006; Todd et al.,

2010). Thus, we propose a similar mechanism in which PSCs release purines (which would be blocked by chelating PSC Ca2+), producing adenosine that activates presynaptic A2ARs and causes synaptic potentiation. Interestingly, the potentiation of synaptic transmission of weak and strong nerve terminals by the A2A agonist suggests that they both remain available for synaptic strengthening. This is consistent with the ‘‘flip-flop’’ concept (Darabid et al., 2014; Walsh and Lichtman, 2003), whereby the weaker of the competing terminal may still overcome the stronger one and occupy the postsynaptic site. This could be a protection mechanism to ensure the innervation of a muscle fiber in case the stronger competing terminal is damaged or no longer active (Turney and Lichtman, 2012). Moreover, the selective potentiation of the strong terminal suggests that PSCs impose the outcome because both inputs possess functional A2ARs. Last, the fact that the effects produced by PSCs Ca2+ chelation and those of A2ARs and P2Y1Rs antagonists are similar, without significant additional effects, strengthens the idea that all elements of the cascade most likely share the same pathway and argues against a possible undesired side effect of Ca2+ chelation in PSCs. Activity-Dependent Synapse Elimination by PSCs A model of glia-mediated modulation of synaptic plasticity is presented in Figure S5. Competing terminals release acetylcholine (ACh) and ATP (Figure S5, 1; Redman and Silinsky, 1994; Smith, 1991), where purines are detected by glial cells trough P2Y1Rs located near the synaptic cleft (Figure S5, 2). Detection of the stronger and weaker terminals results in large and small Ca2+ elevations, respectively, in PSCs (Figure S5, 3), inducing the release of gliotransmitters (purines). This will activate extra-synaptic A2ARs to preferentially potentiate the strongest input (Figure S5, 4). This differential localization of receptors is essential for the distinction between purines released by presynaptic terminals during HFS and purinergic signaling from PSCs that modulates synaptic plasticity (Araque et al., 2014). The selective blockade of Ca2+ elevations in PSCs or the blockade of P2Y1Rs prevented PSCs from detecting neurotransmission and altered plasticity. Furthermore, blockade of P2Y1Rs influenced the outcome of synapse competition. Although the contribution of non-glial P2Y1Rs cannot be completely ruled out using pharmacological approaches, a few elements argue in favor of a PSC-autonomous mechanism. First, immunohistological results suggest that P2Y1Rs do not colocalize with post-synaptic receptors and are more likely located on PSCs, juxtaposed to presynaptic release sites (Darabid et al., 2013). Second, the effects of Ca2+ chelation in PSCs are similar to blockade of P2Y1Rs, suggesting a shared pathway. Finally, PSC Ca2+ activity alone (Ca2+ uncaging) is sufficient to differentially potentiate the release of competing terminals. We showed that daily acute blockade of P2Y1Rs partially and transiently delayed synapse elimination. Our results contrasts with a recent study using a constitutive KO of P2Y1Rs, where no effect on polyinnervation was observed (Heredia et al., 2018). In these mice, receptors were knocked out from all cell types but also throughout development of the nervous system, including the period of fetal and post-natal

Cell Reports 25, 2070–2082, November 20, 2018 2079

development of the NMJ. This could permit early compensation of this blockade by numerous mechanisms involved in the process of synapse formation and maturation (Darabid et al., 2014). These mechanisms can be related to the intrinsic activity of nerve terminals (Buffelli et al., 2002; Favero et al., 2012; Personius and Balice-Gordon, 2001; Personius et al., 2007), the endogenous phagocytic activity of PSCs (Smith et al., 2013; Song et al., 2008), the direct contribution of the muscle fiber (Favero et al., 2009), as well as the availability of trophic factors such as brain derived neurotrophic factor (BDNF) and glial cell line-derived neurotrophic factor (GDNF) (Je et al., 2012, 2013; Keller-Peck et al., 2001a; Nguyen et al., 1998). The presence of multiple mechanisms is consistent with the importance of synapse elimination for survival to ensure proper connectivity of the nervous system. The importance of phagocytic activity of glial cells in synapse elimination is well documented (Chung et al., 2013; Schafer et al., 2012; Stevens et al., 2007). Here we blocked Ca2+-dependent glial activity, not directly targeting the elimination machinery, allowing us to reveal the requirement of the differential detection of neurotransmitter release by PSCs for proper synapse elimination. It would be interesting to determine the relationship between the different regulators of synapse elimination, in particular the detection and regulation of synaptic activity and the ability of PSCs to protect or eliminate nerve terminals. In vivo, P2Y1Rs blockade resulted in a 2-day delay in synapse elimination. The percentage of increase in polyinnervation is among the highest reported in the literature (Brill et al., 2016; Favero et al., 2012; Personius et al., 2008). A delay in synapse elimination rather than a complete blockade following P2Y1R manipulation (Figures 6 and 7) reinforces the idea that many mechanisms work in parallel to ensure the vital process of synapse elimination. Moreover, our results may underestimate the importance of these receptors because the specific antagonist used was easily washable (Figure 4) so that P2Y1Rs might have been blocked for only a short period of time during the day. Altogether, we show that glial cells differentially modulate the activity of competing terminals during synaptic competition at the NMJ, and we present evidence that such a regulation of synaptic activity is important for proper synapse elimination. Thus, we provide insights that may help us to understand the extended role of glial cells in synapse competition and the remodeling of connectivity in the nervous system. STAR+METHODS Detailed methods are provided in the online version of this paper and include the following: d d d d

KEY RESOURCES TABLE CONTACT FOR REAGENT AND RESOURCE SHARING EXPERIMENTAL MODEL AND SUBJECT DETAILS B Mice METHOD DETAILS 2+ B In situ monitoring of PSC Ca responses B In situ monitoring of synaptic activity B In vivo purinergic type 2Y1 receptor blockade

2080 Cell Reports 25, 2070–2082, November 20, 2018

d

QUANTIFICATION AND STATISTICAL ANALYSIS B Quantification B Statistical analysis

SUPPLEMENTAL INFORMATION Supplemental Information includes five figures and one videos and can be found with this article online at https://doi.org/10.1016/j.celrep.2018.10.075. ACKNOWLEDGMENTS The authors wish to thank Dr. Edward Ruthazer for insightful comments. This work was supported by an operating grant to R.R. from the Canadian Institutes of Health Research (MOP-14137) and an NSERC discovery grant, by an equipment grant from the Canadian Foundation of Innovation and by an infrastructure grant from the FRQ-S (Fonds Que´be´cois pour la Recherche – Sante´) to the GRSNC (Groupe de Recherche sur le Syste`me Nerveux Central). H.D. held a studentship from the FRQ-S, and A.S.-P.S. held a studentship from FQR-NT (Fonds Que´be´cois pour la Recherche – Nature & Technologie). AUTHOR CONTRIBUTIONS H.D. and A.S.-P.S. conducted the experiments and analyzed the data. H.D. and R.R. designed the experiments. H.D., A.S.-P.S., and R.R. interpreted the data and wrote the manuscript. DECLARATION OF INTERESTS The authors declare no competing interests. Received: December 13, 2016 Revised: July 23, 2018 Accepted: October 19, 2018 Published: November 20, 2018 REFERENCES Alloisio, S., Cugnoli, C., Ferroni, S., and Nobile, M. (2004). Differential modulation of ATP-induced calcium signalling by A1 and A2 adenosine receptors in cultured cortical astrocytes. Br. J. Pharmacol. 141, 935–942. Araque, A., Carmignoto, G., Haydon, P.G., Oliet, S.H., Robitaille, R., and Volterra, A. (2014). Gliotransmitters travel in time and space. Neuron 81, 728–739. Arbour, D., Tremblay, E., Martineau, E´., Julien, J.-P., and Robitaille, R. (2015). Early and persistent abnormal decoding by glial cells at the neuromuscular junction in an ALS model. J. Neurosci. 35, 688–706. Ataman, B., Ashley, J., Gorczyca, M., Ramachandran, P., Fouquet, W., Sigrist, S.J., and Budnik, V. (2008). Rapid activity-dependent modifications in synaptic structure and function require bidirectional Wnt signaling. Neuron 57, 705–718. Balice-Gordon, R.J., and Lichtman, J.W. (1993). In vivo observations of preand postsynaptic changes during the transition from multiple to single innervation at developing neuromuscular junctions. J. Neurosci. 13, 834–855. Balice-Gordon, R.J., and Lichtman, J.W. (1994). Long-term synapse loss induced by focal blockade of postsynaptic receptors. Nature 372, 519–524. Bialas, A.R., and Stevens, B. (2013). TGF-b signaling regulates neuronal C1q expression and developmental synaptic refinement. Nat. Neurosci. 16, 1773–1782. Bishop, D.L., Misgeld, T., Walsh, M.K., Gan, W.B., and Lichtman, J.W. (2004). Axon branch removal at developing synapses by axosome shedding. Neuron 44, 651–661. Boyer, J.L., Adams, M., Ravi, R.G., Jacobson, K.A., and Harden, T.K. (2002). 2-Chloro N(6)-methyl-(N)-methanocarba-20 -deoxyadenosine-30 ,50 -bisphosphate is a selective high affinity P2Y(1) receptor antagonist. Br. J. Pharmacol. 135, 2004–2010.

Brill, M.S., Kleele, T., Ruschkies, L., Wang, M., Marahori, N.A., Reuter, M.S., Hausrat, T.J., Weigand, E., Fisher, M., Ahles, A., et al. (2016). Branch-Specific Microtubule Destabilization Mediates Axon Branch Loss during Neuromuscular Synapse Elimination. Neuron 92, 845–856. Buffelli, M., Busetto, G., Cangiano, L., and Cangiano, A. (2002). Perinatal switch from synchronous to asynchronous activity of motoneurons: link with synapse elimination. Proc. Natl. Acad. Sci. USA 99, 13200–13205. Buffelli, M., Burgess, R.W., Feng, G., Lobe, C.G., Lichtman, J.W., and Sanes, J.R. (2003). Genetic evidence that relative synaptic efficacy biases the outcome of synaptic competition. Nature 424, 430–434. Busetto, G., Buffelli, M., Tognana, E., Bellico, F., and Cangiano, A. (2000). Hebbian mechanisms revealed by electrical stimulation at developing rat neuromuscular junctions. J. Neurosci. 20, 685–695. Cai, D., Cohen, K.B., Luo, T., Lichtman, J.W., and Sanes, J.R. (2013). Improved tools for the Brainbow toolbox. Nat. Methods 10, 540–547. Campbell, G., and Shatz, C.J. (1992). Synapses formed by identified retinogeniculate axons during the segregation of eye input. J. Neurosci. 12, 1847– 1858. Castonguay, A., and Robitaille, R. (2001). Differential regulation of transmitter release by presynaptic and glial Ca2+ internal stores at the neuromuscular synapse. J. Neurosci. 21, 1911–1922. Chen, C., and Regehr, W.G. (2000). Developmental remodeling of the retinogeniculate synapse. Neuron 28, 955–966. Chung, W.S., Clarke, L.E., Wang, G.X., Stafford, B.K., Sher, A., Chakraborty, C., Joung, J., Foo, L.C., Thompson, A., Chen, C., et al. (2013). Astrocytes mediate synapse elimination through MEGF10 and MERTK pathways. Nature 504, 394–400. Colman, H., Nabekura, J., and Lichtman, J.W. (1997). Alterations in synaptic strength preceding axon withdrawal. Science 275, 356–361. Correia-de-Sa´, P., Timo´teo, M.A., and Ribeiro, J.A. (1996). Presynaptic A1 inhibitory/A2A facilitatory adenosine receptor activation balance depends on motor nerve stimulation paradigm at the rat hemidiaphragm. J. Neurophysiol. 76, 3910–3919. Darabid, H., Arbour, D., and Robitaille, R. (2013). Glial cells decipher synaptic competition at the mammalian neuromuscular junction. J. Neurosci. 33, 1297– 1313. Darabid, H., Perez-Gonzalez, A.P., and Robitaille, R. (2014). Neuromuscular synaptogenesis: coordinating partners with multiple functions. Nat. Rev. Neurosci. 15, 703–718. Del Castillo, J., and Katz, B. (1954). Quantal components of the end-plate potential. J. Physiol. 124, 560–573.

Gorassini, M., Eken, T., Bennett, D.J., Kiehn, O., and Hultborn, H. (2000). Activity of hindlimb motor units during locomotion in the conscious rat. J. Neurophysiol. 83, 2002–2011. Hashimoto, K., and Kano, M. (2013). Synapse elimination in the developing cerebellum. Cell. Mol. Life Sci. 70, 4667–4680. Henneberger, C., Papouin, T., Oliet, S.H., and Rusakov, D.A. (2010). Longterm potentiation depends on release of D-serine from astrocytes. Nature 463, 232–236. Heredia, D.J., Feng, C.Y., Hennig, G.W., Renden, R.B., and Gould, T.W. (2018). Activity-induced Ca2+ signaling in perisynaptic Schwann cells of the early postnatal mouse is mediated by P2Y1 receptors and regulates muscle fatigue. eLife 7, e30839. Hettinger, B.D., Lee, A., Linden, J., and Rosin, D.L. (2001). Ultrastructural localization of adenosine A2A receptors suggests multiple cellular sites for modulation of GABAergic neurons in rat striatum. J. Comp. Neurol. 431, 331–346. Hirata, K., Zhou, C., Nakamura, K., and Kawabuchi, M. (1997). Postnatal development of Schwann cells at neuromuscular junctions, with special reference to synapse elimination. J. Neurocytol. 26, 799–809. Hooks, B.M., and Chen, C. (2006). Distinct roles for spontaneous and visual activity in remodeling of the retinogeniculate synapse. Neuron 52, 281–291. Je, H.S., Yang, F., Ji, Y., Nagappan, G., Hempstead, B.L., and Lu, B. (2012). Role of pro-brain-derived neurotrophic factor (proBDNF) to mature BDNF conversion in activity-dependent competition at developing neuromuscular synapses. Proc. Natl. Acad. Sci. U.S.A. 109, 15924–15929. Je, H.S., Yang, F., Ji, Y., Potluri, S., Fu, X.Q., Luo, Z.G., Nagappan, G., Chan, J.P., Hempstead, B., Son, Y.J., and Lu, B. (2013). ProBDNF and mature BDNF as punishment and reward signals for synapse elimination at mouse neuromuscular junctions. J. Neurosci. 33, 9957–9962. Jourdain, P., Bergersen, L.H., Bhaukaurally, K., Bezzi, P., Santello, M., Domercq, M., Matute, C., Tonello, F., Gundersen, V., and Volterra, A. (2007). Glutamate exocytosis from astrocytes controls synaptic strength. Nat. Neurosci. 10, 331–339. Kamiya, H., and Zucker, R.S. (1994). Residual Ca2+ and short-term synaptic plasticity. Nature 371, 603–606. Katz, L.C., and Shatz, C.J. (1996). Synaptic activity and the construction of cortical circuits. Science 274, 1133–1138. Keller-Peck, C.R., Feng, G., Sanes, J.R., Yan, Q., Lichtman, J.W., and Snider, W.D. (2001a). Glial cell line-derived neurotrophic factor administration in postnatal life results in motor unit enlargement and continuous synaptic remodeling at the neuromuscular junction. J. Neurosci. 21, 6136–6146.

Del Rio, T., and Feller, M.B. (2006). Early retinal activity and visual circuit development. Neuron 52, 221–222.

Keller-Peck, C.R., Walsh, M.K., Gan, W.B., Feng, G., Sanes, J.R., and Lichtman, J.W. (2001b). Asynchronous synapse elimination in neonatal motor units: studies using GFP transgenic mice. Neuron 31, 381–394.

Dunwiddie, T.V., Diao, L., and Proctor, W.R. (1997). Adenine nucleotides undergo rapid, quantitative conversion to adenosine in the extracellular space in rat hippocampus. J. Neurosci. 17, 7673–7682.

Kopp, D.M., Perkel, D.J., and Balice-Gordon, R.J. (2000). Disparity in neurotransmitter release probability among competing inputs during neuromuscular synapse elimination. J. Neurosci. 20, 8771–8779.

Eken, T., Elder, G.C., and Lømo, T. (2008). Development of tonic firing behavior in rat soleus muscle. J. Neurophysiol. 99, 1899–1905.

LaMantia, A.S., and Rakic, P. (1990). Axon overproduction and elimination in the corpus callosum of the developing rhesus monkey. J. Neurosci. 10, 2156–2175.

Favero, M., Massella, O., Cangiano, A., and Buffelli, M. (2009). On the mechanism of action of muscle fibre activity in synapse competition and elimination at the mammalian neuromuscular junction. Eur. J. Neurosci. 29, 2327–2334. Favero, M., Busetto, G., and Cangiano, A. (2012). Spike timing plays a key role in synapse elimination at the neuromuscular junction. Proc. Natl. Acad. Sci. USA 109, E1667–E1675. Fuentes-Medel, Y., Logan, M.A., Ashley, J., Ataman, B., Budnik, V., and Freeman, M.R. (2009). Glia and muscle sculpt neuromuscular arbors by engulfing destabilized synaptic boutons and shed presynaptic debris. PLoS Biol. 7, e1000184. Garcia, N., Priego, M., Obis, T., Santafe, M.M., Toma`s, M., Besalduch, N., Lanuza, M.A., and Toma`s, J. (2013). Adenosine A1 and A2A receptor-mediated modulation of acetylcholine release in the mice neuromuscular junction. Eur. J. Neurosci. 38, 2229–2241.

LeVay, S., Wiesel, T.N., and Hubel, D.H. (1980). The development of ocular dominance columns in normal and visually deprived monkeys. J. Comp. Neurol. 191, 1–51. Lohof, A.M., Delhaye-Bouchaud, N., and Mariani, J. (1996). Synapse elimination in the central nervous system: functional significance and cellular mechanisms. Rev. Neurosci. 7, 85–101. Mariani, J., and Changeux, J.P. (1981). Ontogenesis of olivocerebellar relationships. I. Studies by intracellular recordings of the multiple innervation of Purkinje cells by climbing fibers in the developing rat cerebellum. J. Neurosci. 1, 696–702. Matos, M., Shen, H.Y., Augusto, E., Wang, Y., Wei, C.J., Wang, Y.T., Agostinho, P., Boison, D., Cunha, R.A., and Chen, J.F. (2015). Deletion of adenosine A2A receptors from astrocytes disrupts glutamate homeostasis leading to

Cell Reports 25, 2070–2082, November 20, 2018 2081

psychomotor and cognitive impairment: relevance to schizophrenia. Biol. Psychiatry 78, 763–774.

Ridge, R.M., and Betz, W.J. (1984). The effect of selective, chronic stimulation on motor unit size in developing rat muscle. J. Neurosci. 4, 2614–2620.

Matsuzaki, M., Honkura, N., Ellis-Davies, G.C., and Kasai, H. (2004). Structural basis of long-term potentiation in single dendritic spines. Nature 429, 761–766.

Robitaille, R. (1998). Modulation of synaptic efficacy and synaptic depression by glial cells at the frog neuromuscular junction. Neuron 21, 847–855.

Na¨gerl, U.V., Eberhorn, N., Cambridge, S.B., and Bonhoeffer, T. (2004). Bidirectional activity-dependent morphological plasticity in hippocampal neurons. Neuron 44, 759–767.

Rochon, D., Rousse, I., and Robitaille, R. (2001). Synapse-glia interactions at the mammalian neuromuscular junction. J. Neurosci. 21, 3819–3829.

Navarrete, M., and Araque, A. (2010). Endocannabinoids potentiate synaptic transmission through stimulation of astrocytes. Neuron 68, 113–126. Navarrete, M., Perea, G., Fernandez de Sevilla, D., Go´mez-Gonzalo, M., Nu´n˜ez, A., Martı´n, E.D., and Araque, A. (2012). Astrocytes mediate in vivo cholinergic-induced synaptic plasticity. PLoS Biol. 10, e1001259. Nevian, T., and Helmchen, F. (2007). Calcium indicator loading of neurons using single-cell electroporation. Pflugers Arch. 454, 675–688. Nguyen, Q.T., Parsadanian, A.S., Snider, W.D., and Lichtman, J.W. (1998). Hyperinnervation of neuromuscular junctions caused by GDNF overexpression in muscle. Science 279, 1725–1729. Oliveira, L., Timo´teo, M.A., and Correia-de-Sa´, P. (2004). Tetanic depression is overcome by tonic adenosine A(2A) receptor facilitation of L-type Ca(2+) influx into rat motor nerve terminals. J. Physiol. 560, 157–168. Panatier, A., Theodosis, D.T., Mothet, J.P., Touquet, B., Pollegioni, L., Poulain, D.A., and Oliet, S.H. (2006). Glia-derived D-serine controls NMDA receptor activity and synaptic memory. Cell 125, 775–784. Panatier, A., Valle´e, J., Haber, M., Murai, K.K., Lacaille, J.C., and Robitaille, R. (2011). Astrocytes are endogenous regulators of basal transmission at central synapses. Cell 146, 785–798. Paolicelli, R.C., Bolasco, G., Pagani, F., Maggi, L., Scianni, M., Panzanelli, P., Giustetto, M., Ferreira, T.A., Guiducci, E., Dumas, L., et al. (2011). Synaptic pruning by microglia is necessary for normal brain development. Science 333, 1456–1458. Pascual, O., Casper, K.B., Kubera, C., Zhang, J., Revilla-Sanchez, R., Sul, J.-Y., Takano, H., Moss, S.J., McCarthy, K., and Haydon, P.G. (2005). Astrocytic purinergic signaling coordinates synaptic networks. Science 310, 113–116. Perea, G., and Araque, A. (2007). Astrocytes potentiate transmitter release at single hippocampal synapses. Science 317, 1083–1086. Personius, K.E., and Balice-Gordon, R.J. (2001). Loss of correlated motor neuron activity during synaptic competition at developing neuromuscular synapses. Neuron 31, 395–408. Personius, K.E., Chang, Q., Mentis, G.Z., O’Donovan, M.J., and Balice-Gordon, R.J. (2007). Reduced gap junctional coupling leads to uncorrelated motor neuron firing and precocious neuromuscular synapse elimination. Proc. Natl. Acad. Sci. USA 104, 11808–11813. Personius, K.E., Karnes, J.L., and Parker, S.D. (2008). NMDA receptor blockade maintains correlated motor neuron firing and delays synapse competition at developing neuromuscular junctions. J. Neurosci. 28, 8983– 8992. Rebola, N., Lujan, R., Cunha, R.A., and Mulle, C. (2008). Adenosine A2A receptors are essential for long-term potentiation of NMDA-EPSCs at hippocampal mossy fiber synapses. Neuron 57, 121–134. Redfern, P.A. (1970). Neuromuscular transmission in new-born rats. J. Physiol. 209, 701–709. Redman, R.S., and Silinsky, E.M. (1994). ATP released together with acetylcholine as the mediator of neuromuscular depression at frog motor nerve endings. J. Physiol. 477, 117–127. Ribchester, R.R., and Taxt, T. (1983). Motor unit size and synaptic competition in rat lumbrical muscles reinnervated by active and inactive motor axons. J. Physiol. 344, 89–111.

2082 Cell Reports 25, 2070–2082, November 20, 2018

Santello, M., Bezzi, P., and Volterra, A. (2011). TNFa controls glutamatergic gliotransmission in the hippocampal dentate gyrus. Neuron 69, 988–1001. Schafer, D.P., Lehrman, E.K., Kautzman, A.G., Koyama, R., Mardinly, A.R., Yamasaki, R., Ransohoff, R.M., Greenberg, M.E., Barres, B.A., and Stevens, B. (2012). Microglia sculpt postnatal neural circuits in an activity and complement-dependent manner. Neuron 74, 691–705. Serrano, A., Haddjeri, N., Lacaille, J.C., and Robitaille, R. (2006). GABAergic network activation of glial cells underlies hippocampal heterosynaptic depression. J. Neurosci. 26, 5370–5382. Smith, D.O. (1991). Sources of adenosine released during neuromuscular transmission in the rat. J. Physiol. 432, 343–354. Smith, I.W., Mikesh, M., Lee, Yi., and Thompson, W.J. (2013). Terminal Schwann cells participate in the competition underlying neuromuscular synapse elimination. J. Neurosci. 33, 17724–17736. Song, J.W., Misgeld, T., Kang, H., Knecht, S., Lu, J., Cao, Y., Cotman, S.L., Bishop, D.L., and Lichtman, J.W. (2008). Lysosomal activity associated with developmental axon pruning. J. Neurosci. 28, 8993–9001. Stellwagen, D., and Shatz, C.J. (2002). An instructive role for retinal waves in the development of retinogeniculate connectivity. Neuron 33, 357–367. Stevens, B., Allen, N.J., Vazquez, L.E., Howell, G.R., Christopherson, K.S., Nouri, N., Micheva, K.D., Mehalow, A.K., Huberman, A.D., Stafford, B., et al. (2007). The classical complement cascade mediates CNS synapse elimination. Cell 131, 1164–1178. Tapia, J.C., Wylie, J.D., Kasthuri, N., Hayworth, K.J., Schalek, R., Berger, D.R., Guatimosim, C., Seung, H.S., and Lichtman, J.W. (2012). Pervasive synaptic branch removal in the mammalian neuromuscular system at birth. Neuron 74, 816–829. Todd, K.J., Darabid, H., and Robitaille, R. (2010). Perisynaptic glia discriminate patterns of motor nerve activity and influence plasticity at the neuromuscular junction. J. Neurosci. 30, 11870–11882. Toma`s, J., Santafe´, M.M., Garcia, N., Lanuza, M.A., Toma`s, M., Besalduch, N., Obis, T., Priego, M., and Hurtado, E. (2014). Presynaptic membrane receptors in acetylcholine release modulation in the neuromuscular synapse. J. Neurosci. Res. 92, 543–554. Turney, S.G., and Lichtman, J.W. (2012). Reversing the outcome of synapse elimination at developing neuromuscular junctions in vivo: evidence for synaptic competition and its mechanism. PLoS Biol. 10, e1001352. Walsh, M.K., and Lichtman, J.W. (2003). In vivo time-lapse imaging of synaptic takeover associated with naturally occurring synapse elimination. Neuron 37, 67–73. Wyatt, R.M., and Balice-Gordon, R.J. (2003). Activity-dependent elimination of neuromuscular synapses. J. Neurocytol. 32, 777–794. Yang, J., Yang, H., Liu, Y., Li, X., Qin, L., Lou, H., Duan, S., and Wang, H. (2016). Astrocytes contribute to synapse elimination via type 2 inositol 1,4,5-trisphosphate receptor-dependent release of ATP. eLife 5, e15043. Zhan, Y., Paolicelli, R.C., Sforazzini, F., Weinhard, L., Bolasco, G., Pagani, F., Vyssotski, A.L., Bifone, A., Gozzi, A., Ragozzino, D., and Gross, C.T. (2014). Deficient neuron-microglia signaling results in impaired functional brain connectivity and social behavior. Nat. Neurosci. 17, 400–406.

STAR+METHODS KEY RESOURCES TABLE

REAGENT or RESOURCE

SOURCE

IDENTIFIER

Mouse IgG1, anti-SV2, (concentrate)

Developmental Studies Hybridoma Bank

Cat# SV2c; ID:2315387

Chicken, anti-NFM (neurofilament Medium)

Rockland Antibodies and Assays

Cat# 212-901-D84

Rabbit, anti-S100B

Agilent (DAKO)

Cat# Z0311

Alexa Fluor 647-Affinipure Donkey Anti-Rabbit IgG (H+L) (min X Bov, Ck, Gt, GP, Sy, Hms, Hrs, Hu, Ms, Rat, Shp Sr Prot) (ML)

Jackson Immunoresearch Labs

Cat# 711-605-152

Alexa Fluor 488 AffiniPure Goat Anti-Mouse IgG, Fcg subclass 1 specific (min X Hu, Bov, Rb Sr Prot)

Jackson Immunoresearch Labs

Cat# 115-545-205

Alexa Fluor 488 AffiniPure Donkey Anti-Chicken IgY (IgG) (H+L); (min X Bov, Gt, GP, Sy Hms, Hrs, Hu, Ms, Rb, Rat, Shp Sr Prot)

Jackson Immunoresearch Labs

Cat# 703-545-155

Antibodies

Chemicals, Peptides, and Recombinant Proteins Adenosine 5-tri disodium salt hydratephosphate

Millipore Sigma

Cat# A2383

Alexa Fluor 594 hydrazide, sodium salt 1mg

ThermoFisher Scientific

Cat# A10438

Alpha-bungarotoxin, Alexa Fluor 594 conjugate

ThermoFisher Scientific

Cat# B13423

ATP disodium salt

Millipore Sigma

Cat# A2383

CGS 21680 hydrochloride

Tocris

Cat# 1063

d-tubocurarine hydrochloride pentahydrate

Millipore Sigma

Cat# T2379

Diazo-2 tetrapotassium salt *cell impermeant*

ThermoFisher Scientific

Cat# D3034

Fluo-4 pentapotassium salt, *cell impermeant*

ThermoFisher Scientific

Cat# F14200

Ketamine hydrochloride, sterile (100mg/ml)

Comparative Medecine Animal Resource center, McGill University // Narketan

Cat# 440893

Magnesium Chloride hexahydrate (Rees Ringer Solution)

Millipore Sigma

Cat# M2670

Magnesium Sulfate heptahydrate (mofidied Rees Ringer Solution) with low Ca2+ (1mM) / high Mg2+ (6-7mM)

Millipore Sigma

Cat# M1880

MRS 2179

ABCAM

Cat# AB120415

Normal Donkey serum

Jackson Immunoresearch

Cat# 017-000-121

o-nitrophenyl EGTA, tetrapotassium salt (NP-EGTA) *cell impermeant*

ThermoFisher Scientific

Cat# N6802

Prolong Gold Antifade Mountant with DAPI

ThermoFisher Scientific

Cat# P36935

SCH-58261

Millipore Sigma

Cat# S4568

Xylazine sterile injection (20 mg/ml) -

CDMV

Cat#3770

184 Sylgard (3.9 kg kit)

Paisley Products of Canada Inc

Cat#AVDC00184036

Charles River Canada https://www.criver.com/ sites/default/files/resources/CD1IGSMouseModelInformationSheet.pdf

Strain number #022 nomenclature: Crl:CD1(ICR))

Borosilicate glass with filament, OD:1,2mm / ID: 0,94mm/ 10 cm (for local application of drug)

Sutter Instrument

Cat# BF120-94-10

Borosilicate glass capillaries, 1,00mm OD (for electrophysiology recording and single cell electroporation)

WPI

Cat# 1B100F-4

Experimental Models: Organisms/Strains CD-1 IGS mice; age: 7-8 days old

Other

Cell Reports 25, 2070–2082.e1–e6, November 20, 2018 e1

CONTACT FOR REAGENT AND RESOURCE SHARING Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Dr. Richard Robitaille ([email protected]). EXPERIMENTAL MODEL AND SUBJECT DETAILS Mice All animal care and experiments were performed in accordance with the guidelines of the Canadian Council of Animal Care (CCAC) and the Comite´ de de´ontologie de l’expe´rimentation sur les animaux (CDEA) from Universite´ de Montre´al. Commercial CD-1 IGS mice (Charles River Canada) were used for all experiments. An adult female mouse with a litter of 10 male pups aged between P2-P5 were bought together. They were maintained in their own housing unit under controlled temperature (22 C) and illumination (12 h dark/light cycle), until the pups were used at P7-P8 (for in situ experiments) or at most P14 (for in vivo experiments). Mice had ad libitum access to food and fresh water. METHOD DETAILS In situ monitoring of PSC Ca2+ responses General procedure All animals used were P7–P8 male CD-1 mice. The procedures referenced here are all described in detail below. First, nerve-muscle preparations were obtained, pinned in a recording chamber and installed under a confocal microscope. Once the preparation was stably perfused, all visible PSCs from different NMJs were electroporated. In some experiments, P2Y1R antagonist treatment was applied prior to PSC activation. PSCs were activated by local ATP or adenosine application or high frequency stimulation (HFS) and their Ca2+ responses were concomitantly recorded using confocal imaging. While several PSCs at several NMJs per muscle were studied by local applications of ATP or adenosine, only one NMJ per muscle was studied in experiments where the activation of PSCs was done using HFS. Nerve-muscle preparations Mice were anesthetized by a lethal intraperitoneal injection of ketamine (300 g/kg) / Xylazine (20 g/kg). The Soleus muscle was dissected under oxygenated (95% O2, 5% CO2) Rees solution (in mM): 110 NaCl, 5 KCl, 1 MgCl2, 25 NaHCO3, 2 CaCl2, 11 glucose, 0.3 glutamate, 0.4 glutamine, 5 BES, 4.34X107 cocarboxylase and 0.036 choline chloride. The pH of the oxygenated solution was 7.3. The preparation was pinned in a 184 Sylgard-coated recording chamber (Paisley Products of Canada). When the motor nerve stimulations were required, the Soleus muscle was dissected with its innervation up to the ventral roots (L3-L5). Two or three ventral root segments were obtained and independently stimulated by three suction-stimulation electrodes (square pulses; 0.1 mV to 2.0 V, 0.1 ms duration) using a Master-8 stimulator (AMPI). This allowed the independent stimulation of two inputs competing at the same NMJ with axons located in distinct stimulation electrodes as described previously (Darabid et al., 2013; Kopp et al., 2000). The suction-stimulation electrodes were made of polyethylene (PE-160) tubes containing 0.125 mm platinum wire (WPI) electrodes. The tip of the tube was stretched so that the size of its opening would match the diameter of the corresponding ventral root segment. A reference platinum electrode was located close to the tube opening to allow optimal current flow upon stimulation. When experiments required that the relative synaptic strength of inputs to be determined, intracellular recordings of the synaptic activity of competing inputs was recorded prior to PSC electroporation using the procedure described in a following section (In situ monitoring of synaptic activity). Extracellular solution Once installed on the experimental platform under the confocal microscope, preparations were perfused using the same Rees extracellular solution described for nerve-muscle dissection. However, muscle contractions were prevented by partially blocking postsynaptic nicotinic acetylcholine receptors (nAChRs) with D-tubocurarine chloride (2.0-3.5 mM, Millipore Sigma). The experiments were performed at 28 C-30 C and the temperature of the solution was constantly monitored and automatically adjusted using a TC-324B solution heater (Warner Instruments). Single cell electroporation of PSCs A pipette (10-12 MU) was pulled from borosilicate glass tubes (o.d. 1.2 mm) using a P97 micropipette puller from Sutter Instrument. It was filled with 500 mM of the fluorescent indicator Alexa Fluor 594 (ThermoFisher Scientific) and 800 mM of the Ca2+-indicator Fluo-4 Pentapotassium Salt (ThermoFisher Scientific) diluted in the extracellular Rees solution. The pipette was mounted on a pipette holder with a 0.125 mm platinum stimulating wire connected to a Master-8 stimulator (AMPI). The reference electrode was placed in the bath. PSCs were easily identified with transmitted light microscopy. The pipette was approached to the soma of a PSC using a micromanipulator under visual guidance. The tip of the pipette was positioned close to the PSC soma but did not touch the cell or any tissue. One to three single negative square pulses (15 V, 10 ms) were applied to open the cell membrane (Darabid et al., 2013; Nevian and Helmchen, 2007). The procedure was repeated to load other visible PSCs at the same NMJ (Figure S1).. In addition, we previously confirmed that the electroporated cells are S100B-postive PSCs (Darabid et al., 2013). The preparation was allowed to rest for at least 20 min prior further manipulations.

e2 Cell Reports 25, 2070–2082.e1–e6, November 20, 2018

Pharmacological treatment In some experiments, the purinergic type 2Y1 receptors (P2Y1Rs) antagonist MRS2179 (20 mM; Abcam) was added to the extracellular solution at least 20 min prior to PSC activation. High frequency stimulation In several experiments, Ca2+ responses in PSCs were monitored in response to HFS (30 s at 50 Hz; square pulses of 0.1 mV to 2.0 V and a duration of 0.1 ms) of terminals competing at the same NMJ. First, one terminal was stimulated using the described parameters while monitoring Ca2+ responses in PSCs. Then, after a rest period of 20 min, the second terminal was stimulated using the same procedure. ATP or adenosine local application In some experiments, Ca2+ responses in PSCs were assessed following ATP (5 or 10 mM, Millipore Sigma) or adenosine (10 mM, Millipore Sigma) local applications from a glass pipette (5-8 MU) positioned at proximity of cells. Positive pressure pulses (15 PSI, 150 ms) with a Picospritzer II (Parker Instruments) were used to locally apply the agonists. They were diluted in the extracellular solution used for perfusion. PSC Ca2+-imaging PSCs activity was measured using Ca2+-imaging during the experimental activation and monitored with an Olympus FV1000 microscope using a 60X water-immersion lens (0.90 NA; Olympus). The 488 and 594 nm laser lines were used for the excitation of Fluo-4 and Alexa Fluor 594 respectively. The emitted fluorescence was detected using the multispectral detection feature (bandpass filters: 500-545 nm to detect the green Ca2+ indicator and 570-670 nm for the red dye). Exclusion criteria Any preparation that showed signs of damage were excluded. Such signs include swollen muscle fibers, low muscle fiber resting membrane potential, unstable recordings, uncharacteristic morphology of PSCs, non-responsive PSCs to ATP in control conditions or non-specific Ca2+-rises in PSCs. In situ monitoring of synaptic activity General procedure All animals used were P7–P8 male CD-1 mice. The procedures referenced here are all described in detail below. First, nerve-muscle preparations were obtained, pinned in a recording chamber and installed under a confocal microscope. Once the preparation was stably perfused in a low Ca2+/high Mg2+ solution, intracellular recordings were performed in muscle fibers in order to determine the relative synaptic strength of the competing inputs. The same recording was kept for the entire duration of the experiment. The extracellular solution was then changed to one containing normal levels of Ca2+ and Mg2+, but supplemented with D-tubocurarine chloride to partially block postsynaptic nicotinic receptors and prevent muscle contractions. Only dually-innervated NMJs were studied. They showed EPPs evoked by the independent stimulation of two of the three ventral roots. If an increase in EPP amplitude (EPP steps) occurred while increasing the stimulation intensity for a given suction electrode, the recording was discarded as this would suggest that multiple axons innervating the same NMJs were present within the ventral root stimulated by that stimulation electrode (Buffelli et al., 2002; Busetto et al., 2000; Redfern, 1970). This was done after the transition to D-tubocurarine chloride containing extracellular solution. In some experiments, specific receptor antagonists (P2Y1R or A2AR) were also added to the extracellular solution at this stage. When required by the experimental design, all visible PSCs at the target NMJ where electroporated during the extracellular solution transition (20 min). Once this transition was complete, the baseline synaptic activity of both inputs was recorded before any experimental manipulation (HFS, A2A agonist application, Diazo2 or NP-EGTA photoactivation). The effect of the experimental manipulation on synaptic activity of competing inputs was then recorded for at least 20 min. In some experiments, the Ca2+ activity of PSCs was recorded while the experimental manipulation was delivered (HFS, Diazo2 and NP-EGTA). Unless otherwise stated in the main text, only one NMJ per muscle was studied. Furthermore, only one competing input was analyzed per muscle to study the synaptic plasticity and avoid potential bias due to the sequence of stimulated competing terminals (i.e., strong input first then the weak or vice-versa). Nerve-muscle preparations Nerve-muscle preparations were obtained as described in the In situ monitoring of PSC Ca2+ responses section. However, the Soleus muscle was always dissected with its innervation up to the ventral roots (L3-L5). Intracellular recordings of synaptic activity Synaptic events were recorded using an Axoclamp 2B amplifier (Axon Instruments) and further amplified (100X) and filtered at 2 kHz by a Warner Instruments amplifier. EPPs were digitized at 10 kHz with DigiData 1322A (Axon Instruments). Data were collected and analyzed using pClamp 8.0 software (Axon Instruments). Endplate potentials (EPPs) were evoked by nerve stimulation (square pulses of 0.2 mV to 2.0 V; 0.1 ms duration) using a Master-8 stimulator (AMPI). Synaptic recordings we performed using sharp intracellular electrodes (40-60 MU filled with 3M KCl) pulled from a borosilicate glass tubes (o.d. 1.0 mm) using a P97 micropipette puller from Sutter Instrument. The internal solution of the electrode was connected to the headstage with a 0.125 mm silver wire.

Cell Reports 25, 2070–2082.e1–e6, November 20, 2018 e3

All recordings of synaptic strength were obtained using a modified Ringer’s solution with low Ca2+ (1 mM)/ high Mg2+ (6–7 mM), which also blocks muscle contractions. The ventral roots were stimulated at a frequency of 0.2 Hz with an intensity that was twice the threshold for eliciting EPPs. The method used to determine the synaptic strength of each input is described in the section Quantification and statistical analysis. Following extracellular solution transition, the baseline of synaptic activity was recorded for one of the two inputs by stimulating the corresponding ventral root at a frequency of 0.2 Hz for at least 10 min of stable EPP amplitude recording. Following experimental activation, EPPs were recorded at a frequency of 0.2 Hz for at least 20 min to monitor changes in synaptic transmission. Extracellular solution transition Rees solution with normal Ca2+/Mg2+ concentration was perfused after recording synaptic activity of each input using low Ca2+/ high Mg2+ solution to determine the synaptic strength. The solution was perfused for at least 20 min before baseline synaptic activity was recorded and was used for the remainder of the experiment. In this condition, muscle contractions were prevented by partially blocking postsynaptic nicotinic acetylcholine receptors (nAChRs) with D-tubocurarine chloride (2.0-3.5 mM, Millipore Sigma). In some experiments, the purinergic type 2Y1 receptors (P2Y1Rs) antagonist MRS2179 (20 mM; Abcam) or adenosine type 2A receptors (A2ARs) antagonist SCH58261 (100 nM; Millipore Sigma) was added to the extracellular solution. Single cell electroporation of PSCs Single cell electroporation of PSCs was done as presented in the previous section (In situ monitoring of PSC Ca2+ responses). In specific sets of experiments, either 2 mM Diazo2 (ThermoFisher Scientific) or 2 mM NP-EGTA (ThermoFisher Scientific) was added to the electroporation solution. Synaptic activity was constantly monitored by intracellular recordings from the muscle fiber to ensure no changes in EPPs amplitude and thus no undesired effect of the single-cell electroporation technique. In all cases, the preparation was allowed to rest for at least 20 min before further manipulations. Photoactivation of Diazo2 and NP-EGTA A 405 nm laser was used to photoactivate Diazo2 (multiple pulses: 7% power, 1 s ON/1 s OFF for 20 s; Figure S3) or NP-EGTA (single pulse: 7% power, 500 ms). The photoactivation region was positioned to cover all PSCs somata and processes. The Olympus FV1000 SIM Lightpath and tornado function was used to allow fast and efficient photoactivation. Photobleaching was monitored and compared to the red Alexa Fluor 594 channel. Possible photodamage was constantly controlled by monitoring synaptic activity, cell morphology as well as the general aspect of PSCs and muscle fibers. Adenosine type 2A receptors agonist In some experiments, the adenosine type 2A receptors (A2ARs) agonist CGS21680 (7 nM; Tocris) was added to the extracellular solution following baseline synaptic activity recording and used as the activating stimulus. High frequency stimulation In some experiments, HFS (30 s at 50 Hz; square pulses of 0.1 mV to 2.0 V and duration of 0.1 ms) of one of the competing terminals was used as the activating stimulus. PSC Ca2+-imaging PSCs activity was measured as described in the In situ monitoring of PSC Ca2+ responses section. Exclusion criteria Recordings which showed a resting membrane potentials lower than 50 mV, changes of more than 10% of the amplitude of the EPP from either inputs during the baseline recording where discarded. Experiments where recordings were lost before 20 min following any manipulation were also discarded. In vivo purinergic type 2Y1 receptor blockade Daily injections Two individual work spaces were prepared and cleaned, one for the injection and the other for the post-injection recovery. Seringes (BD Safety Glide Insulin, 3/10ml 31G x 5/16 TW) containing the drug (MRS2179; 40 mM in sterile 0.9% NaCl) or the 0.9% NaCl sterile solution were prepared. Half of the cage containing the mother and pups was placed on a heated mat. Another heating pad was prepared with a clean underpad impregnated with the litter’s scent (to minimize stress) for the awakening of the injected pup. A clean pair of gloves to manipulate the pups was also impregnated with the litter’s scent. One at the time, pups were anesthetized on an iceblock covered with an underpad impregnated with the litter’s scent. After 1-2 min, or as soon as the pup was still, 0.02 mL to 0.05 mL of the drug was slowly administered subcutaneously next to the Soleus muscle in one leg. The contralateral leg was then immediately injected with 0.02-0.05 mL of the 0.9% NaCl sterile solution. Warm and sterile 0.9% NaCl (0.2-0.5 ml) was also injected subcutaneously on each flank of the pup to help recovery. After the injections, the pup was quickly moved to the prepared heating pad until it started moving normally. It was then returned to the mother. The procedure lasted about 2 min. To minimize muscle and tissue damage, alternate daily injection between the outside and inside of the injected leg were performed. Once the pups were all injected, they were kept under surveillance for at least 2 hr before being returned to the animal facility. This allowed us to ensure that the pups were not harmed during the procedure and that they were groomed and licked appropriately before being returned to their housing unit. Pups and mother were checked daily. The procedure was repeated daily until the pups were sacrificed at P8, P10, P12 or P14. Muscles from MRS2179 and saline-injected legs were dissected for immunohistochemistry staining.

e4 Cell Reports 25, 2070–2082.e1–e6, November 20, 2018

End points Cannibalism rarely happens if the procedure is respected and if time of separation between pup and mother is minimized. In the event of cannibalism, the entire litter was euthanized. If a limb inflammation following chronic treatment was suspected, unless it subsided quickly, the injections were stopped and the entire litter was euthanized. Immunohistochemistry of NMJs Immunohistochemical labeling was performed as described previously (Darabid et al., 2013). Soleus muscles were dissected using the same dissection solution as described in the In situ monitoring of PSC Ca2+ responses section and pinned in a Sylgard-coated dish. Then, they were fixed for 10 min in 4% formaldehyde at room temperature and permeabilized in 100% cold methanol for 6 min at 20 C. Nonspecific labeling was minimized by incubation in 10% normal donkey serum (NDS) and 0.01% Triton X-100 solution for 20 min. All primary and secondary antibodies were prepared in a solution of PBS containing 0.01% Triton-X and 2% NDS. Preparations were incubated overnight at 4 C with a rabbit anti-S100B (1:250, Agilent). Muscles were then incubated in chicken anti-NF-M (1:1000; Rockland Antibodies and Assays Antibodies and Assays) and mouse IgG1 anti-SV2 (1:1500; Developmental Studies Hybridoma Bank) antibodies for 90 min. Preparations were incubated with Alexa Fluor 488 donkey anti-chicken, Alexa Fluor 647 donkey anti-rabbit and DL488 goat anti-mouse IgG1 secondary antibodies (all at 1:500) for 60 min at room temperature. Finally, muscles were incubated with a conjugated Alexa Fluor 594 a-bungarotoxin (0.75 mg/ml) for 45 min. After each step, muscles were washed in PBS containing 0.01% Triton X-100 (3 times, 5 min). The preparations were then mounted in the Prolong Gold antifade mounting medium containing DAPI (ThermoFisher Scientific) and all labels were observed in two phases and using the spectral detection feature of an Olympus FV1000 confocal microscope with a 63x oil-immersion lens (1.43 NA; Olympus). An approximate airy disk value of 1 was obtained by adjusting the Pinhole. Images were not manipulated after the acquisition. Exclusion criteria Any preparation that showed signs of damage such as uncharacteristic nAchR distribution or aberrant nerve terminals morphology were excluded. Muscles with non-specific or atypical staining were also discarded. QUANTIFICATION AND STATISTICAL ANALYSIS Quantification PSC Ca2+-imaging PSCs at the NMJ area were imaged at a rate of 2.1 Hz using the imaging software included with the FV1000 Olympus microscope (Fluoview software Ver 4.02). Ca2+ changes reflected in fluorescence (Fl) intensity were analyzed over each PSC soma and expressed as ððFl­FlRest Þ=FlRest Þ 3 100 : EPP measurements Data were collected and analyzed using pClamp 8.0 software (Axon Instruments). EPP amplitude is expressed as % of amplitude change compared to baseline: ðEPP amplitude in mV=mean baseline EPP amplitude in mVÞ 3 100 : Each point represents the mean amplitude of 24 EPPs (2 min). Synaptic strength and facilitation The synaptic strength of each competing terminal was determined by calculating the quantal content as m = Loge ð#nerve impulses=#failuresÞ (Del Castillo and Katz, 1954) and the paired pulse facilitation (Darabid et al., 2013; Kopp et al., 2000). Two stimuli were given at 10 ms interval to determine the paired pulse facilitation, which was calculated as  F = 2nd EPP amplitude 1st EPP amplitudeðincluding failuresÞ By definition, terminals’ were identified as weak or strong based on their respective values of quantal content. All analyses, including the calculations of synaptic strength, were performed offline once experiments were completed. Therefore, the sequence for the stimulation of competing inputs (i.e., strong input first then the weak or vice-versa) was blind at the time of the experiment and unbiased. Polyinnervation quantification The state of polyinnervation was evaluated by counting the number of independent inputs (labeled by anti-neurofilament-M (NF-M) and anti-synaptic vesicular protein 2 (SV2) antibodies) that contacted a single NMJ endplate (nAChR area labeled by a-bungarotoxin). Each NMJ was then classified as mono- or poly-innervated. At least 20 surface NMJs were analyzed per muscle. Data from MRS2179-injected muscles were grouped together and compared to saline-injected ones.

Cell Reports 25, 2070–2082.e1–e6, November 20, 2018 e5

Statistical analysis Results are presented as mean ± SEM n represents the number of PSCs and N the number of muscles in Ca2+-imaging experiments. N represents the number of recorded NMJs in electrophysiological experiments. N represents the number of muscles in the in vivo injection experiments. Unless stated otherwise, only one PSC/NMJ was kept for further analyses and only one dually-innervated NMJ was recorded per muscle. Paired t tests were performed when comparing PSC Ca2+-responses induced by the two competing terminals during the same experiment. When data were found not to conform to normality using a D’Agostino and Pearson omnibus normality test, Mann– Whitney U tests were used. Unpaired t tests were performed to compare two different conditions from different experiments. One-way ANOVA, repeated-measures, with Tukey’s multiple comparison tests were used to compare three groups or more. Analyses were deemed significant at p < 0.05. All statistical analyses were performed using the GraphPad Prism 5.01 software. All analyses details (mean ± SEM, biological replicate number, p value, statistical test) are reported in the relevant location of the Results section. Relevant biological replicate numbers are also stated in the Figure legends.

e6 Cell Reports 25, 2070–2082.e1–e6, November 20, 2018