Neurexins cluster Ca²⁺ channels within the presynaptic active zone

Abstract

To achieve ultrafast neurotransmission, neurons assemble synapses with highly organized presynaptic and postsynaptic nanomachines that are aligned by synaptic adhesion molecules. How functional assembly of presynaptic active zones is controlled via trans‐synaptic interactions remains unknown. Here, we conditionally deleted all three neurexin adhesion molecules from presynaptic neurons of the calyx of Held in the mouse auditory system, a model synapse that allows precise biophysical analyses of synaptic properties. The pan‐neurexin deletion had no effect on synapse development or the basic release machinery, but dramatically impaired fast neurotransmitter release. The overall properties of presynaptic calcium ion channels appeared normal, as reflected by the similar characteristics of calcium currents recorded at the nerve terminals. However, the pan‐neurexin deletion significantly impaired the tight coupling of calcium influx to exocytosis, thereby suppressing neurotransmitter release. Furthermore, the pan‐neurexin deletion reduced the function of calcium‐activated BK potassium channels, whose activation depends on their tight association with presynaptic calcium channels. Together, these results suggest that neurexins perform a major function at the calyx synapse in coupling presynaptic calcium channels to release sites.

Mammalian neurexins play a major function at the calyx synapse in organizing presynaptic calcium channels at neurotransmitter release sites.

Introduction

An action potential initiates synaptic transmission by opening voltage‐gated Ca²⁺‐channels that are positioned in the presynaptic active zone in close proximity to neurotransmitter release sites (Kaeser & Regehr, 2014; Biederer et al, 2017; Chanaday & Kavalali, 2018; Walter et al, 2018). The inflowing Ca²⁺ then triggers neurotransmitter release by binding to synaptotagmins (Sudhof, 2013). Neurotransmitter release is fast and transient because Ca²⁺‐channels are closely coupled to release‐ready synaptic vesicles, enabling rapid Ca²⁺‐triggering of release and rapid removal of the inflowing Ca²⁺ ions (Eggermann et al, 2011; Schneggenburger et al, 2012; Sudhof, 2012). Ca²⁺‐channels are tethered to the active zone by binding to protein complexes containing RIMs and RIM‐binding proteins that interact directly with Ca²⁺‐channels and with the release machinery (Han et al, 2011; Kaeser et al, 2011; Liu et al, 2011; Acuna et al, 2015, 2016; Li & Kavalali, 2015; Bohme et al, 2016). How the various components of active zones are assembled precisely opposite to postsynaptic specializations, however, remains unclear.

Neurexins are evolutionarily conserved presynaptic adhesion molecules (reviewed in ref. Sudhof, 2017). Accumulating studies have provided compelling evidence for a central role of neurexins in shaping synapse properties, but arrive at diverse conclusions about what neurexins actually do. Most genetic deletions of neurexins in mice do not lead to a loss of synapse numbers, but often cause decreases in the numbers of a subset of synapses (Missler et al, 2003; Aoto et al, 2015; Chen et al, 2017; Zhang et al, 2018). Moreover, neurexins were found to enable the function of presynaptic Ca²⁺‐channels (Missler et al, 2003; Chen et al, 2017) and overexpression of neurexin‐1 (Nrxn1) with Ca_V2.1 Ca²⁺‐channels increased their currents (Brockhaus et al, 2018), but the specific role for neurexins in either gating Ca²⁺‐channel function or in Ca²⁺‐channel trafficking remained elusive. One potential problem is that few studies on vertebrate neurexins examined their functions in terms of the precise properties of a synapse because such properties are difficult to measure directly in most synapses. Pioneering studies in Caenorhabditis elegans (C. elegans) uncovered a role for presynaptic neurexins in synapse assembly via interactions with postsynaptic neuroligins (Maro et al, 2015; Tu et al, 2015) and revealed a role for neurexins in synaptogenesis that was redundant with that of the wnt‐receptor frizzled (Kurshan et al, 2018). However, at a different synaptic interaction between neurons in C. elegans, neurexins were shown to be postsynaptic and neuroligin deemed to be presynaptic (Hu et al, 2012), and at yet a third pair of synaptic connections between neurons in C. elegans, neurexins were found to be essential for the plasticity of mature, sexually dimorphic neurons (Hart & Hobert, 2018). A more uniform picture of neurexins emerged from work on the Drosophila neuromuscular junction, at which presynaptic neurexin was found to be essential for the correct active zone apposition (Li et al, 2007). Here, neurexin interacted intracellularly with the WAVE complex and spinophilin, and extracellularly with neuroligins to organize the architecture of the synapse (Muhammad et al, 2015; Banerjee et al, 2017). However, even in Drosophila, additional functions for neurexins in organizing dendritic arborizations (D'Rozario et al, 2016; Liu et al, 2017) or even in retinoid transport (Tian et al, 2013) were described, raising questions about what neurexins actually do.

Insight into neurexin functions is not only important for understanding synapses, but also for understanding neuropsychiatric diseases. Mutations in all neurexin genes were associated with a variety of such disorders. NRXN1 mutations are the most frequent, with NRXN1 copy‐number variations among the most often observed single‐gene mutations in schizophrenia and Tourette syndrome (Huang et al, 2017; Marshall et al, 2017; Kasem et al, 2018; Yuan et al, 2018; Al Shehhi et al, 2019). The broad spectrum of clinical presentations associated with neurexin mutations in human patients is consistent with a general role of neurexins in synapses, but how neurexin mutations predispose to neuropsychiatric diseases remains unclear. Analysis of human neurons carrying NRXN1 mutations revealed a synaptic impairment, but owing to the technical limitations inherent in studies of cultured neurons, the precise nature of this impairment remained elusive (Pak et al, 2015).

Thus, despite extensive work, a coherent validated concept of the functions of neurexins—apart from a generally accepted role as a synaptic organizer—remains elusive, and their contribution to neuropsychiatric diseases is poorly understood. Two problems in particular have hindered progress. First, the expression of three neurexin genes makes studies of neurexins in vertebrates difficult because of the potential for redundancy. Complicating the multiplicity of genes, vertebrate neurexins are additionally diversified by the presence of multiple promoters and by extensive alternative splicing (Ullrich et al, 1995; Schreiner et al, 2014; Treutlein et al, 2014). This diversity is functionally important for the multiple promoter use (Anderson et al, 2015; Aoto et al, 2015) and at least for one site of alternative splicing, SS#4 (Aoto et al, 2013; Dai et al, 2019). Thus, for assessment of general functions of neurexins it is necessary to ablate expression of all neurexins. Second, analysis of neurexin function likely requires precise measurements of synaptic transmission at a defined synapse, which is not possible in most systems because of the inability to patch nerve terminals and characterize their properties. Most vertebrate synapses and all invertebrate synapses are subject to this limitation. To overcome these technical challenges, we have here used conditional triple KO mice that allow deletion of all vertebrate neurexins to analyze the effect of a pan‐neurexin deletion on the properties of a defined synapse. In our analyses, we focused on the calyx of Held synapse because of its accessibility to high‐resolution measurements of vesicle exocytosis and of presynaptic Ca²⁺‐currents (Takahashi, 2015; Baydyuk et al, 2016; Korber & Kuner, 2016; Joris & Trussell, 2018; Sakaba, 2018). Our results demonstrate that neurexins are required at this central synapse for the overall organization of the active zone, in particular for the precise clustering of Ca²⁺‐channels that are tightly coupled to the release machinery and of BK‐channels. However, we find that neurexins are not essential for the initial establishment and maintenance of the synapse, the release machinery itself, or Ca²⁺‐channel function as such. Thus at the calyx synapse, neurexins perform a surprisingly discrete essential role in the functional assembly of presynaptic Ca²⁺‐channels in active zones.

Results

Deletion of all neurexins does not decrease synapse numbers at the calyx of Held

To analyze the overall function of neurexins in synapses that enables a precise biophysical dissection of synaptic transmission and its modulation by pre‐ and postsynaptic signals, we focused on the calyx of Held synapse, a glutamatergic synapse in the medial nucleus of the trapezoid body (MNTB) in the auditory brainstem. In the calyx synapse, a large presynaptic terminal containing hundreds of synaptic junctions wraps around the soma of a principal neuron, enabling precise measurements of pre‐ and postsynaptic mechanisms with an unparalleled resolution (Fig 1A; Takahashi, 2015; Baydyuk et al, 2016; Korber & Kuner, 2016; Joris & Trussell, 2018; Sakaba, 2018). We bred conditional Nrxn123 triple KO mice that enable deletion of all neurexins (Chen et al, 2017) with Pv‐Cre mice that express Cre‐recombinase under control of the parvalbumin gene (Hippenmeyer et al, 2005; Fig 1B). Pv‐Cre mice exhibit robust cre‐recombination in excitatory presynaptic neurons of the calyx synapse on postnatal day 4 (Zhang et al, 2017), making it a useful Cre‐driver line for presynaptic molecule manipulations. We generated conditional Nrxn123 triple KO mice without and with a heterozygous Pv‐Cre allele, enabling comparisons between littermate conditional Nrxn123 triple KO mice lacking (referred to as control) or containing Cre‐expression (referred to as Nrxn123 TKO mice) in presynaptic neurons forming calyx synapses.

Figure 1. Neurexins are not required for synapse formation at the calyx of Held.

• A
  Diagram of the calyx of Held synapse.
• B
  Strategy for selective deletion of all neurexins at the calyx of Held by crossing PV‐Cre driver mice with neurexin‐1/2/3 (Nrxn123) triple conditional knockout (TKO) mice (Zhang et al, 2017; Chen et al, 2017).
• C
  Representative images of calyx synapses from littermate control and Nrxn123 TKO mice. Brainstem sections were labeled with antibodies to vGluT1 (red) and Syt2 (green). Scale bar, 10 μm.
• D
  Summary graphs of the synaptic vGluT1 and Syt2 immunostaining intensity (normalized to control).
• E–H
  Representative traces of spontaneous EPSCs (sEPSCs) (E), and summary graphs of the sEPSC frequency (F), amplitude (G), and kinetics (H) recorded from littermate control and Nrxn123 TKO mice.

Data information: Data in (D, F–H) are means ± SEM with individual data points. Number of image sections/mice (D) or of cells from at least three mice per group (F–H) analyzed are indicated in the bars. No statistical differences were found by Student's t‐test.

We first examined whether deletion of all neurexins from calyx synapses may impact formation or maintenance of synaptic junctions. Immunolabeling of MNTB sections from control and Nrxn123 TKO mice did not uncover any changes in the staining intensity of the presynaptic vesicle proteins vGluT1 (specific for excitatory synapses), vGAT (specific for inhibitory synapses), and synaptotagmin‐2 (Syt2, present primarily in excitatory synapses), or of the excitatory postsynaptic AMPA‐receptor GluA1 (Figs 1C and D, and EV1A–D). Moreover, measurements of synaptic activity by monitoring spontaneous excitatory postsynaptic currents (EPSCs) in acute brainstem slices did not reveal changes in the frequency, amplitude, or kinetics of individual synaptic events (Fig 1E–H). Moreover, protein quantifications of the dissected MNTB failed to uncover a change in the levels of key synaptic protein components (Fig EV2). Thus, neurexins by themselves are not essential for the initial formation or maintenance of calyx synapses.

Figure EV1. Neurexins are not required for synapse formation at the calyx of Held as evidenced by a lack of change in the signals for presynaptic vesicle markers (vGluT1 and vGAT) or postsynaptic AMPA‐type glutamate receptors (GluA1) after complete deletion of all neurexins in conditional Nrxn123 triple KO mice (TKO) using a Pv‐Cre driver line (related to Fig 1).

1. Representative brainstem sections containing the MNTB from littermate conditional Nrxn123 TKO mice that either do not express Cre‐recombinase (control), or that harbor a parvalbumin‐Cre (Pv‐Cre) allele driving expression of Cre‐recombinase in both pre‐ and postsynaptic neurons of the calyx of Held synapse (Nrxn123 TKO; see Zhang et al, 2017). Sections were labeled with antibodies to vGluT1 (red) and GluA1 (green). Scale bar, 10 μm.
2. Summary of vGluT1 and GluA1 immunostaining intensity (normalized to control).
3. Brainstem sections containing the MNTB from control and Nrxn123 TKO mice that were labeled with antibodies to vGluT1 (red) and vGAT (green). Scale bar, 10 μm.
4. Summary of vGluT1 and vGAT immunostaining intensity (normalized to control).

Data information: Data in (B) and (D) are means ± SEM. Number of image sections/mice analyzed are indicated in the bars (B and D); no statistically significant differences were observed as assessed by Student's t‐test.

Figure EV2. Deletion of all neurexins using Pv‐Cre driver mice has no major impact on the expression of key synaptic proteins in the dissected MNTB (related to Fig 1).

1. Diagram of dissecting MNTB nucleus for immunoblot analysis.
2. Example images of immunoblots of synaptic proteins.
3. Summary graphs of protein levels (normalized to Tuji of control mice).

Data information: Data are means ± SEM (n = 4 mice/group); no statistically significant differences were observed as assessed by Student's t‐test.

Deletion of all neurexins impairs evoked synaptic transmission at the calyx of Held

We next recorded evoked EPSCs at the calyx of Held synapse using afferent fiber stimulation in acute slices. We applied a paired stimulus with a 10‐ms interval every 20 s (Fig 2A) with a bath solution containing 2 mM Ca²⁺, which is the standard Ca²⁺‐concentration used in slice recordings but likely unphysiologically high, since the normal extracellular Ca²⁺‐concentration in vivo is 1.2 mM (Forsberg et al, 2019).

Figure 2. Pan‐neurexin deletion severely impairs evoked synaptic transmission at calyx synapses.

• A
  Representative traces of EPSCs evoked by paired stimuli separated by 10 ms and repeated every 20 s, recorded in a standard bath solution containing 2 mM Ca²⁺.
• B
  Pan‐neurexin deletion suppresses synaptic transmission. Summary graphs show the amplitude (left) and charge transfer (right) of the first EPSC recorded in response to the paired stimuli.
• C
  Pan‐neurexin deletion more than doubles the paired‐pulse ratio (PPR), suggesting a large decrease in release probability.
• D
  Pan‐neurexin deletion decelerates the EPSC time course. Summary graphs show the rise time (left) and decay time constants (right) of the first EPSC in response to the paired stimuli.
• E, F
  Pan‐neurexin deletion does not significantly alter the cumulative EPSC amplitude during a high‐frequency stimulus train (100 Hz for 0.5 s). Left, representative ESPC traces; right, cumulative summary plot of the EPSC amplitudes during the train (dotted lines show linear regression fits for estimating the cumulative EPSC amplitude by back‐extrapolation to zero time, which is used to correct for vesicle replenishment during the train).
• G
  Pan‐neurexin deletion does not decrease the readily releasable pool of vesicles, but lowers the initial release probability during a high‐frequency stimulus train. Summary graphs show the cumulative EPSC amplitudes extrapolated to time zero as an estimate of the readily releasable pool size (left), and the ratio of the first EPSC amplitude divided by the cumulative EPSC amplitudes extrapolated to time zero as an estimate of the initial release probability (right).

Data information: Data are means ± SEM with individual data points. Number of cells (from at least three mice per group) analyzed are indicated in the bars (B, D, and G) or in the graph (F) and apply to all graphs in a series; statistical differences were assessed by Student's t‐test (**P < 0.01;***P < 0.001; non‐significant differences are not marked).

The pan‐neurexin deletion caused a dramatic decrease in the peak amplitude (~60% decrease) and total charge transfer of EPSCs (~50% decrease) (Fig 2B). In addition, the pan‐neurexin deletion produced a significant increase in the paired‐pulse ratio (~115%; Fig 2C) and a modest but significant elevation of EPSC rise times (~20%; Fig 2D). This phenotype suggests a decreased release probability, which could be due to a decrease in the readily releasable pool of vesicles, the Ca²⁺‐sensitivity of release, or action potential‐induced Ca²⁺‐influx.

To differentiate among these possibilities, we applied high‐frequency stimulus trains (100 Hz for 0.5 s; Fig 2E). Although the pan‐neurexin deletion decreased the starting EPSC amplitude, it did not significantly decrease the cumulative EPSC amplitude during the train (Fig 2F and 2G). This result suggests that the neurotransmitter release machinery and Ca²⁺‐sensor for exocytosis were intact and that specifically the readily releasable pool of vesicles was not impaired. As expected, the fraction of the first EPSC amplitude divided by the cumulative EPSC was decreased, confirming an impaired release probability (Fig 2G). In addition to EPSC measurements, we monitored evoked IPSCs, but detected no change in Nrxn123 TKO mice, presumably because the Pv‐Cre driver line does not support Cre‐expression in presynaptic inhibitory neurons in the MNTB (Fig EV3A–D). Moreover, we examined the effect of the pan‐neurexin deletion on action potential generation and properties in excitatory presynaptic calyx terminals, but again could not detect any neurexin‐dependent change (Fig EV4A–C).

Figure EV3. Deletion of all neurexins in conditional Nrxn123 TKO mice using a Pv‐Cre driver line has no impact on inhibitory synaptic inputs onto MNTB neurons, presumably because Pv‐Cre drives expression only in postsynaptic MNTB neurons but not in the inhibitory presynaptic input neurons (as opposed to excitatory presynaptic calyx neurons; see Zhang et al, 2017; related to Fig 2).

1. At P12–14, inhibitory synapses on the MNTB are primarily glycinergic. Inhibitory postsynaptic currents (IPSCs) evoked by afferent fiber stimulation were recorded from MNTB neurons in acute brainstem slices from control mice at P12–P14, with sequential addition of 25 μM strychnine (Str) and 50 μM picrotoxin (PTX). Left, example traces of IPSCs; right, summary graph of the percent inhibition of the IPSC amplitude after sequential addition of strychnine and picrotoxin.
2. Sample traces of IPSCs from littermate control and Pv‐Cre conditional Nrxn123 TKO mice at P12–P14.
3. Summary graphs of the amplitude and integrated charge of IPSCs.
4. Summary graphs of the rise times and decay time constants of IPSCs.

Data information: Data in (A, C, and D) are means ± SEM. Number of cells (from three mice per group) analyzed are indicated in the bars (A, C and D); no statistically significant differences were observed as assessed by Student's t‐test.

Figure EV4. Deletion of neurexins has no major effect on presynaptic action potentials measured in patched terminals of calyx of Held synapses (related to Fig 2).

1. Example traces of presynaptic action potentials in littermate control and Nrxn123 TKO mice monitored in presynaptic calyx terminals.
2. Summary graphs of the action potential amplitude and duration. Data are means ± SEM. Number of cells (from three mice per group) analyzed are indicated in the bars; no statistically significant differences were observed as assessed by Student's t‐test
3. Example traces of presynaptic APs firing at 100 Hz, showing that there are no action potential failures.

The decreased release probability but normal readily releasable pool and functional Ca²⁺‐sensing mechanism in neurexin‐deficient calyx synapses suggest that the pan‐neurexin deletion may affect the function of presynaptic Ca²⁺‐channels in calyx terminals or the coupling of the Ca²⁺‐channels to the release machinery. To test this hypothesis, we measured EPSCs with a reduced (1 mM) extracellular Ca²⁺‐concentration in the bath solution (Fig 3A–D), or with that standard (2 mM) extracellular Ca²⁺‐concentration in the bath but with brain slice pretreated with 0.2 mM EGTA‐AM, which is a cell‐permeable Ca²⁺‐chelator (Fig 3E–H). Both dramatically aggravated the pan‐neurexin deletion phenotype. Specifically, the 1 mM Ca²⁺‐concentration nearly completely suppressed the EPSC amplitude and synaptic charge transfer (> 97% and > 95% decrease, respectively; Fig 3B), more than doubled the paired‐pulse ratio (~250% increase; Fig 3C), and decelerated both the EPSC rise and decay kinetics (> 300% and ~70% increase, respectively; Fig 3D). Since this is a more physiological condition than the standard 2 mM Ca²⁺‐concentration, the pan‐neurexin deletion likely ablates synaptic transmission under physiological conditions. Similarly, EGTA‐AM pretreatment also severely suppressed the EPSCs, increased the paired‐pulse ratio, and decelerated the EPSC time course, confirming that either the magnitude or distribution of presynaptic Ca²⁺‐influx are impaired by the pan‐neurexin deletion (Fig 3E–H).

Figure 3. Pan‐neurexin deletion renders high sensitivity of evoked synaptic transmission to reduced Ca²⁺‐signals at calyx synapses.

• A–D
  Same as Fig 2A–D, except that ACSF contained only 1 mM Ca²⁺, which is close to the estimated physiological concentration of extracellular Ca²⁺ (Forsberg et al, 2019).
• E–H
  Same as Fig 2A–D, except that the slice was incubated with 0.2 mM EGTA‐AM for 30 min before recording, which introduces EGTA into the presynaptic terminals.

Data information: Data are means ± SEM with individual data points. Number of cells (from at least three mice per group) analyzed are indicated in the bars (C, G) and apply to all graphs in a series; statistical differences were assessed by Student's t‐test (**P < 0.01;***P < 0.001; non‐significant differences are not marked).

Viewed together, these data demonstrate that neurexins are essential at the calyx synapse for triggering neurotransmitter release by an action potential. The fact that the pan‐neurexin deletion increases the paired‐pulse ratio suggests that the pan‐neurexin deletion impairs the release probability, which is consistent with the decreased EPSC kinetics. The observation that lowering the extracellular Ca²⁺‐concentration to 1 mM or buffering intracellular Ca²⁺ with the slow Ca²⁺‐chelator EGTA causes an almost complete block of release indicates that action potential‐induced Ca²⁺‐influx is impaired, either directly or by decreasing the coupling of Ca²⁺‐influx to release. The finding that the pan‐neurexin deletion has no significant effect on the cumulative EPSC amplitude during a stimulus train indicates that the Ca²⁺‐triggered release machinery is intact since the accumulating Ca²⁺ during the train appears fully able to trigger neurotransmitter release, although it does not rule out a change in the Ca²⁺‐sensitivity of the release machinery. To further analyze the pan‐neurexin deletion phenotype, we thus proceeded to measure presynaptic Ca²⁺‐currents and vesicle exocytosis directly in the nerve terminals.

The pan‐neurexin deletion does not impair Ca²⁺‐channel function, but destroys tight coupling of Ca²⁺‐channels to the release machinery

Neurexins have previously been shown to enable the function of presynaptic Ca²⁺‐channels in other synapses (Missler et al, 2003; Chen et al, 2017; Brockhaus et al, 2018). To directly test whether neurexins are involved in controlling Ca²⁺‐channel properties or numbers at the calyx of Held synapse, we performed whole‐cell patch‐clamped recording to measure presynaptic Ca²⁺‐currents at the calyx terminal. The terminals were maintained at a −80 mV holding potential and sequentially depolarized from −60 mV to + 40 mV in 10 mV steps. No significant differences between control and neurexin‐deficient terminals were found in the maximal Ca²⁺ current amplitude, the maximal Ca²⁺ current density, or the voltage dependence of Ca²⁺ currents (Fig 4A and B). In addition, Ca²⁺ current activation and deactivation kinetics (measured by depolarization to 10 mV) were not changed by pan‐neurexin deletion (Fig 4C). The amplitude and charge of Ca²⁺ currents induced by action potential‐equivalent depolarizations were the same (Fig 4D). Finally, the predominant contribution of P/Q‐type Ca²⁺‐channels to the total Ca²⁺ current remained unaffected after deletion of all neurexins as 200 nM agatoxin blocked presynaptic Ca²⁺ currents similarly in control and Nrxn123 TKO mice (Fig 4E and F). In contrast to previous reports, unexpectedly, our data demonstrate that neurexins do not control the overall properties or expression of presynaptic Ca²⁺‐channels at the central synapse.

Figure 4. Deletion of all neurexins from the calyx of Held synapse does not alter the size or properties of presynaptic Ca²⁺‐currents.

1. Example traces of presynaptic Ca²⁺‐currents recorded in patched calyx terminals as induced by a 50‐ms step depolarization from −60 mV to +40 mV in 10 mV increments at a holding potential of −80 mV.
2. Summary graphs of the I–V relationship for the peak Ca²⁺‐current amplitude (left) and of the peak Ca²⁺‐current density (right).
3. Example traces (left) and summary graphs (right) of the Ca²⁺‐current activation and deactivation kinetics. Currents were induced by a 50‐ms step depolarization from −80 mV to +10 mV.
4. Example traces of Ca²⁺‐currents evoked by an action potential‐equivalent depolarization (AP_e, from −80 mV to +17 mV for 1 ms; left), and summary graph of the AP_e‐evoked Ca²⁺‐current peak amplitude and charge (right).
5. Example traces of presynaptic Ca²⁺‐currents induced by a 50‐ms step depolarization from −60 mV to +40 mV in 10 mV increments at a holding potential of −80 mV before and after perfusion of 200 nM agatoxin (Agtx), a selective P/Q‐type Ca²⁺‐channel blocker.
6. Summary graph of the relative contribution of P/Q‐type Ca²⁺‐channels to the total presynaptic Ca²⁺‐currents, as quantified by the relative reduction in Ca²⁺‐currents evoked by step depolarization to +10 mV, in control and Nrxn123 TKO synapses.

Data information: Data in (B–D, F) are means ± SEM. Number of cells (from three mice per group) analyzed are indicated in the graph (B) or in the bars (C, D, and F); no statistically significant differences were observed as assessed by Student's t‐test.

To further explore whether the coupling between Ca²⁺‐channels and transmitter release machinery may be impacted by pan‐neurexin deletion, which could effectively change the Ca²⁺ signals near release site, we measured simultaneously the presynaptic Ca²⁺‐current and synaptic vesicle exocytosis (monitored via capacitance measurements; Lindau & Neher, 1988; Sun & Wu, 2001; Fig 5A). We compared these recordings in calyx terminals from littermate control and Nrxn123 TKO mice and examined the effects of increasing concentrations of the slow Ca²⁺‐chelator EGTA (directly transferred into the patched terminal via the patch pipette) on the presynaptic Ca²⁺‐currents and capacitance change. Since EGTA binds Ca²⁺ relatively slowly, EGTA introduced into a nerve terminal selectively inhibits Ca²⁺‐triggered exocytosis of synaptic vesicles that are more distant from Ca²⁺‐channels (Eggermann & Jonas, 2011). Thus, increases in the distance between Ca²⁺‐channels and release‐ready synaptic vesicles at the active zone will enhance EGTA‐dependent inhibition of Ca²⁺‐triggered vesicle exocytosis.

Figure 5. Pan‐neurexin deletion renders depolarization‐induced synaptic vesicle exocytosis sensitive to the slow Ca²⁺‐chelator EGTA without affecting total Ca²⁺‐influx into the nerve terminal, Ca²⁺‐triggering of release, or the size of the readily releasable pool.

• A
  Diagram of direct simultaneous measurements of the presynaptic capacitance jump (C_m) and Ca²⁺‐currents (I_Ca) evoked by depolarization of a patched nerve terminal.
• B
  Example traces of the presynaptic capacitance and Ca²⁺‐currents at the calyx terminals induced by a 20‐ms step depolarization from −80 mV to +10 mV with an extracellular solution containing 2 mM Ca²⁺, and an internal pipette solution containing a standard 0.05 mM BAPTA.
• C
  Summary graphs of the depolarization‐induced capacitance jump that monitors synaptic vesicle exocytosis (left), and the peak Ca²⁺‐current density (right).
• D, E
  Same as (B, C), except that 0.5 mM EGTA was added in addition to 0.05 mM BAPTA to the patch pipette internal solution.
• F, G
  Same as (B, C), except that 1.0 mM EGTA was added in addition to 0.05 mM BAPTA to the patch pipette internal solution.
• H, I
  Same as (B, C), except that 10.0 mM EGTA was added in addition to 0.05 mM BAPTA to the patch pipette internal solution.
• J, K
  Summary plots of the relationship between the total capacitance jump (J) or between the Ca²⁺‐current density (K) and the concentration of EGTA in the patch pipette.

Data information: Data are means ± SEM with individual data points. Number of cells (from at least three mice per group) analyzed are indicated in the bars (C, E, G, and I) or shown in the graph (J and K); statistical differences were assessed by Student's t‐test (*P < 0.05; **P < 0.01).

In the absence of EGTA, the pan‐neurexin deletion had no effect on either the total Ca²⁺‐current or the total Ca²⁺‐triggered capacitance change, demonstrating that the pan‐neurexin deletion neither decreased the overall number of functional Ca²⁺‐channels in the presynaptic terminal nor lowered the size of the readily releasable pool of vesicles (Fig 5B and C). Introduction of 0.5 mM EGTA into the presynaptic cytosol had no significant effect on the Ca²⁺‐current and caused only a ~10% decrease in the depolarization‐induced capacitance jump in control terminals, suggesting that most vesicles in normal terminals are close to Ca²⁺‐channels (Fig 5D and E). Again, the pan‐neurexin deletion had no effect on the Ca²⁺‐current, but it now caused a ~45% decrease in the depolarization‐induced capacitance jump compared to the control, revealing a change in the relative position of Ca²⁺‐channels to release‐ready vesicles without a change in total Ca²⁺‐channel function (Fig 5D and E). The inhibition of exocytosis by EGTA was enhanced by increasing EGTA concentrations, but leveled off at ~40% of control synapses and ~60% for pan‐neurexin‐deficient synapses, suggesting that even in pan‐neurexin‐deficient synapses a significant portion of release‐ready vesicles are still close to Ca²⁺‐channels (Fig 5F). These results suggest that the pan‐neurexin deletion disrupts the spatial coupling of Ca²⁺‐channels with synaptic vesicles in the presynaptic active zone without affecting either the total number of Ca²⁺‐channels in the nerve terminal or the release machinery itself.

The pan‐neurexin deletion impairs BK‐channel function at the calyx of Held

The impairment in the tight coupling between Ca²⁺‐channels and synaptic vesicles may suggest either a disorganization of Ca²⁺‐channels in presynaptic active zones or defects in synaptic vesicle tethering/priming. BK‐channels are enriched and tightly coupled with Ca²⁺‐channels at presynaptic active zones (Contet et al, 2016; Griguoli et al, 2016). The recruitment of BK‐channels to the active zone has been shown to be mediated, at least in part, by the direct binding of BK‐channels to the active zone protein RIM‐BP (Sclip et al, 2018). We argued that disorganization of Ca²⁺ channels per se may also impact on the functioning of BK‐channels. To test this hypothesis, we directly patched the terminals and measured the Ca²⁺‐currents and BK‐currents after blocking voltage‐gated Na⁺ and K⁺ channels with TTX and 4‐AP (Fig 6A). Bath application of iberiotoxin (IbTx, 200 nM), a specific inhibitor of BK‐channels, suppressed the majority of the 4‐AP insensitive K⁺ currents that correspond to BK‐currents (Fig 6B). We calculated the BK‐currents by subtracting IBTX insensitive K⁺ currents from the total 4‐AP insensitive K⁺ currents (Fig 6C). Interestingly, we found a striking reduction in BK‐current density (~75% decrease; Fig 6D) in the presynaptic terminal of Nrxn123 TKO mice, as compared to control littermate. The Ca²⁺‐current density was unchanged as described in Fig 5, as was the overall terminal capacitance, suggesting that the fine organization but not the overall structure of the calyx terminal was altered.

Figure 6. Pan‐neurexin deletion depletes K⁺‐currents carried by presynaptic BK‐channels in the calyx of Held synapse.

1. Representative traces of Ca²⁺‐currents and 4‐AP insensitive K⁺‐currents evoked by step depolarizations (from −50 mV to +10 mV in 10 mV increments), recorded from the calyx terminals in acute slices from littermate control and Nrxn123 TKO mice at P12–P14.
2. Same as (A) but after addition of 200 nM iberiotoxin (IbTx) by perfusion.
3. Representative traces of presynaptic BK‐currents, which are calculated by subtracting currents shown in (B) from currents shown in (A).
4. Summary graphs of the capacitance, Ca²⁺‐current density, and BK‐current density.

Data information: Data are means ± SEM with individual data points. Number of cells (from at least three mice per group) analyzed are indicated in the bars (D); Statistical differences were assessed by Student's t‐test (**P < 0.01).

The pan‐neurexin deletion causes a partial loss of P/Q‐type Ca²⁺‐channels from the presynaptic active zone

To visualize a possible dissociation or loss of P/Q‐type Ca²⁺‐channels, which dominate in controlling transmitter release from presynaptic active zones of the calyx of Held, caused by deletion of all neurexins, we stained brainstem cryosections from littermate control and Nrxn123 TKO mice with antibodies to vGluT1, Ca_V2.1 Ca²⁺‐channels, and Bassoon, an active zone‐specific molecule (Fig 7A). We then quantified the vGluT1, Ca_V2.1 Ca²⁺‐channel, and Bassoon signals. Since Ca_V2.1 Ca²⁺‐channels are also present postsynaptically and outside of the presynaptic plasma membrane (although in less enriched form), we quantified their signal only in the presynaptic area stained for vGluT1 (Fig 7B) and measured the size of that area separately as a further control (Fig 7C).

Figure 7. Pan‐neurexin deletion depletes Ca_V2.1‐type Ca²⁺‐channels and Bassoon from the presynaptic active zone of the calyx of Held.

1. Representative images of brainstem cryosections stained by triple immunofluorescence labeling for vGluT1 (green), Ca_V2.1‐type Ca²⁺‐channels (red), and the active zone protein Bassoon (purple). Cryosections were obtained from littermate control and Nrxn123 TKO mice at P12–14. Scale bar, 10 μm.
2. Summary graphs of the vGluT1, Ca_V2.1, and Bassoon immunostaining intensity (normalized to control; Ca_V2.1 and Bassoon signals were quantified over the vGluT1‐positive area to measure active zone‐localized proteins).
3. Summary graphs of the diameter and total size of the vGluT1‐positive presynaptic area of the calyx of Held.

Data information: Data are means ± SEM with individual data points. Number of brain sections and mice analyzed are indicated in the bars (B and C); statistical differences were assessed by Student's t‐test (***P < 0.001).

We found that the pan‐neurexin TKO decreased the signal of Ca_V2.1 Ca²⁺‐channels at the presynaptic area nearly 50% and lowered the Bassoon signal ~25% (Figs 7B and EV5). Strikingly, the vGluT1‐positive area was unchanged in size and shape (Fig 7C). Thus, the pan‐neurexin deletion causes a loss of active zone protein Bassoon and Ca_V2.1 Ca²⁺‐channel protein at the terminals. The decrease in Bassoon signal may further suggest that the deletion of neurexins could induce other changes in the active zone and that the function of neurexins in the calyx synapse is not restricted to recruiting Ca²⁺‐channels, but may broadly extend to other organizational features of the active zone.

Figure EV5. Deletion of all neurexins modestly but significantly reduces the levels of the active zone protein Bassoon in the presynaptic terminal of the calyx of Held synapse (related to Fig 7).

1. Representative brainstem sections containing the MNTB from littermate control and Nrxn123 TKO mice, stained with antibodies to vGluT1 (green) and Bassoon (red). Scale bar, 10 μm.
2. Summary graph of vGluT1 and Bassoon immunostaining intensity (normalized to control).

Data information: Data are means ± SEM. Number of image sections/mice analyzed are indicated in the bars; whereas no statistically significant differences were observed for the vGluT1 staining intensity as assessed by Student's t‐test, the Nrxn123 TKO did induce a significant decrease in Bassoon staining intensity (**P < 0.01).

Discussion

Here, we have examined the overall function of neurexins by a biophysical analysis of the phenotype produced by deleting all neurexins in the calyx of Held synapse, arguably the best central synapse for high‐resolution analyses of synaptic transmission (Takahashi, 2015; Baydyuk et al, 2016; Korber & Kuner, 2016; Joris & Trussell, 2018; Sakaba, 2018). Our results show that neurexins profoundly control the properties of calyx synapses by organizing their active zones, even though they are not directly required for assembly of either the synapses themselves or their release machinery. Specifically, our data suggest three major conclusions.

First, neurexins are not essential in the calyx synapse for the establishment or the maintenance of the synaptic junction, or for the assembly of the release apparatus. The evidence for this conclusion is that deletion of all neurexins did not decrease the density of synaptic junctions in the calyx of Held synapse, as evidenced by immunocytochemistry for synaptic vesicle markers and by measurements of spontaneous synaptic events (Fig 1). Moreover, the pan‐neurexin deletion did not impair neurotransmitter release when such release was induced by high concentrations of Ca²⁺ during high‐frequency stimulus trains (Fig 2E–G) or during prolonged depolarizations (Fig 5A–C).

Second, neurexins enable normal neurotransmitter release induced by an action potential in the calyx synapse, such that under physiological Ca²⁺‐concentrations, the pan‐neurexin deletion nearly abolishes release. This conclusion is based on the observation that action potential‐induced neurotransmitter release was severely impaired by the pan‐neurexin deletion at an extracellular Ca²⁺‐concentration of 2 mM (Fig 2A–D), and almost completely ablated (> 95% inhibition) by the pan‐neurexin deletion at an extracellular Ca²⁺‐concentration 1 mM Ca²⁺ (Fig 3A–D). Since ACSF contains 2 mM Ca²⁺ and the ionized Ca²⁺‐concentration in the brain's extracellular space is 1 mM, the 1 mM Ca²⁺‐concentration represents a more physiological Ca²⁺‐concentration (Forsberg et al, 2019).

Third, neurexins are essential for action potential‐induced neurotransmitter release by organizing the tight spatial coupling of Ca²⁺‐channels to release sites in the active zone of the calyx synapse. The pan‐neurexin deletion did not affect subtype expression or function of Ca²⁺‐channels as such, but reduced effective Ca²⁺ signal reaching the Ca²⁺‐sensors for release. The evidence for this conclusion is provided by the findings that introduction of EGTA into the nerve terminal blocked release even when Ca²⁺‐influx was maximal (Figs 3E–H and 5), that direct measurements of presynaptic Ca²⁺‐influx showed that the pan‐neurexin‐deletion caused no change in Ca²⁺‐channel function (Figs 4 and 5), and that the pan‐neurexin deletion reduced CaV_2.1 Ca²⁺‐channels from the active zone as judged by immunocytochemistry (Fig 7). By organizing the nano‐domain coupling of Ca²⁺‐channels to neurotransmitter release sites that is known to be a critical feature of synapses (Eggermann et al, 2011; Schneggenburger et al, 2012; Walter et al, 2018), neurexins perform a crucial role in enabling the precision and speed of synaptic transmission.

Our data confirm previous findings that neurexins are important for presynaptic Ca²⁺‐channel function (Missler et al, 2003; Zhang et al, 2005; Yamashita et al, 2013; Chen et al, 2017; Tong et al, 2017; Brockhaus et al, 2018), but none of the previous studies defined the actual nature of the observed impairments because direct measurements of presynaptic Ca²⁺‐fluxes are only possible in calyx synapses. By directly measuring Ca²⁺‐channel function in patched presynaptic calyx terminals and simultaneously monitoring synaptic vesicle exocytosis, we could show that the pan‐neurexin deletion does not affect voltage‐gated Ca²⁺‐channels as such, but selectively impairs the coupling of Ca²⁺‐influx to neurotransmitter release (Figs 3, 4, 5). This provides the first direct description of the relation of neurexin function to the organization of presynaptic Ca²⁺‐channels.

Our data further suggest that at least at the calyx synapse, neurexins are general organizers of active zone assembly that confer onto presynaptic terminals modulation by BK‐channels without affecting the release machinery and the initial formation and maintenance of synapses directly. It is clear that neurexins perform different functions at diverse synapses, possibly owing to differences in what neurexin isoforms are specifically expressed in a given presynaptic neuron and which neurexin ligands are presented by the corresponding postsynaptic neuron. We hypothesize that the neurexin functions we uncovered here are also controlled by trans‐synaptic ligands presented by postsynaptic neurons. The functions we describe now explain our previous observations in cortical synapses formed by inhibitory somatostatin‐positive interneurons (Chen et al, 2017), suggesting that they are shared by many synapses. However, these functions are clearly only one facet of many activities performed by neurexins, as evidenced for example by the dramatic trans‐synaptic regulation of postsynaptic AMPA and NMDA receptors mediated by presynaptic Nrxn1 and Nrxn3 alternative splicing in the CA1→subiculum synapses (Dai et al, 2019).

Major new questions now arise. First, how do neurexins organize active zone assembly? The similar phenotypes in terms of Ca²⁺‐channel coupling and BK‐channels observed with the pan‐neurexin deletions here and the RIM‐BP deletions in previous studies (Acuna et al, 2015; Sclip et al, 2018) suggest an interaction of neurexins with presynaptic active zone components. Such interactions might be mediated by CASK (Hata et al, 1996), but the precise molecular pathways remain to be characterized. Second, our pan‐neurexin deletions ablated neurexin expression postnatally, but it is possible that neurexins also have a developmentally earlier function in synapses. Detailed studies ruled out a role for neurexins in axonal pathfinding in mice (Dudanova et al, 2007), although it is possible that such functions operate in invertebrates (D'Rozario et al, 2016; Liu et al, 2017; Hart & Hobert, 2018). Although constitutive deletions of neurexins provided little evidence for an essential role for neurexins in the initial establishment of synapses (Missler et al, 2003), such a role was uncovered in C. elegans when neurexins were deleted together with the wnt‐receptor frizzled (Kurshan et al, 2018). Thus, it is possible that neurexins perform a role in the initial establishment of synapses that is either no longer operative postnatally or rendered redundant by other molecules, a possibility that needs to be tested using more complex genetic experiments. Third, how is neurexin‐dependent assembly of active zones regulated? Is this assembly dynamic and subject to control by presynaptic receptors, and is it controlled by trans‐synaptic signaling via postsynaptic ligands? These questions are crucial for a general understanding of how synapse properties are specified, and how synaptic plasticity shapes such properties. Again, detailed molecular and genetic approaches will be required to address these important questions.

Materials and Methods

Mouse breeding, genotyping, and husbandry

All experiments were approved by the Institutional Animal Care and Use Committee at Stanford University. All experiments were performed using littermates of either sex. Triple Nrxn123 conditional KO mice were described previously (Chen et al, 2017) and were crossed with PV‐IRES‐Cre driver line (Hippenmeyer et al, 2005) to generate cell‐specific Nrxn123 deletion at the calyx of Held synapse. No statistical tests were used to predetermine sample size because the effect size was not known before the experiments, and determining the effect size would have required as many mice as the actual experiments. All electrophysiology experiments were performed blindly without knowledge by the experimenter of the mouse genotypes. All experiments were performed on mice at P12–14.

The primer sequences used for genotyping were as follows:

Nrxn1 flox	5′ GTAGCCTGTTTACTGCAGTTCATTCC 3′ and
	5′ CAAGCACAGGATGTAATGGCCTTTC 3′
Nrxn2 flox	5′ CAGGGTAGGGTGTGGAATGAGGTC 3′ and
	5′ GTTGAGCCTCACATCCCATTTGTCT 3′
Nrxn3 flox	5′ AATAGCAGAGGGGTGTGACAC 3′ and
	5′ CGTGGGGTATTTACGGATGAG 3′
Cre	5′ GAACCTGATGGACATGTTCAGG 3′ and
	5′ AGTGCGTTCGAACGCTAGAGCCTGT 3′

Preparation of brain slices for electrophysiology

Coronal brain slices containing the MNTB nucleus were prepared as described previously (Sun et al, 2007). In brief, mice of postnatal days 12–14 were decapitated; brains were rapidly isolated and glued on the cutting chamber of a vibratome (VT1200s; Leica), which was immersed in oxygenated cold ACSF containing (in mM): 119 NaCl, 26 NaHCO₃, 10 glucose, 1.25 NaH₂PO₄, 2.5 KCl, 0.05 CaCl₂, 3 MgCl₂, 2 Na‐pyruvate, and 0.5 ascorbic acid, pH 7.4. Transverse 160–200 μm slices were sectioned and transferred into a beaker with bubbled ACSF containing (in mM): 119 NaCl, 26 NaHCO₃, 10 glucose, 1.25 NaH₂PO₄, 2.5 KCl, 2 CaCl₂, 1 MgCl₂, 2 Na‐pyruvate, and 0.5 ascorbic acid, pH 7.4. After recovery at 35°C for 45 min, slices were stored at room temperature (< 21–23°C) for experiments.

Postsynaptic voltage‐clamp recordings

Whole‐cell voltage‐clamp recordings were made from principal cells in the MNTB visualized by infrared differential interference contrast (IR‐DIC) video microscopy (Axioskop 2; Zeiss). Patch‐clamp recording was made with the EPC 10 amplifier (HEKA, Lambrecht, Germany). Patch pipettes (resistance of 3–4 MΩ) were pulled using borosilicate glass (WPI) on a two‐stage vertical puller (Narishige). Series resistances (< 12 MΩ) were compensated by 70–90% in order to maintain a residual resistance of < 2 MΩ.

For EPSC recordings, the MNTB cells were voltage‐clamped at −70 mV, and EPSCs were recorded in ACSF. Picrotoxin (100 μM), strychnine (2 μM), and D‐AP5 (50 μM) were routinely added to block GABA_A receptors, glycine receptors, and NMDA receptors, respectively. The pipette internal solution contained the following (in mM): 120 Cs‐gluconate, 20 tetraethylammonium‐Cl, 20 HEPES, 2 EGTA, 4 MgATP, 0.4 NaGTP, 10 phosphocreatine, and 2 Qx‐314. Afferent fiber stimulation (0.1–0.5 mA, 0.1 ms) was delivered using a bipolar electrode positioned halfway between midline and the MNTB to ensure that the calyx was activated in an all‐or‐none manner.

For IPSC recordings, the cells at MNTB were voltage‐clamped at −70 mV, and IPSCs were recorded in ACSF containing CNQX (20 μM) and D‐AP5 (50 μM) to block AMPA receptors and NMDA receptors, respectively. The internal solution contained the following (in mM): 75 CsCl, 68 Cs‐gluconate, 10 HEPES, 2 EGTA, 4 MgATP, 0.4 NaGTP, and 10 phosphocreatine.

Presynaptic Ca²⁺‐current and capacitance measurements

To directly record Ca²⁺‐current and vesicle release from the presynaptic terminal, the calyces were patched and voltage‐clamped in whole‐cell mode (Sun & Wu, 2001). Capacitance measurements were achieved using the lock‐in program of PatchMaster (HEKA, Lambrecht, Germany) together with the EPC 10 amplifier (HEKA, Lambrecht, Germany). The frequency of the sinusoidal stimulus was set at 1,000 Hz, and the peak‐to‐peak amplitude was less than 60 mV. The patched terminals were routinely held at the resting membrane potential of −80 mV. Ca²⁺ influx and vesicle fusion were evoked by depolarization of the cell from the holding membrane potential to +10 mV for 20 ms except specified otherwise. The membrane capacitance was not measured during stimulus. The capacitance jump evoked by 20‐ms depolarization was calculated as the capacitance difference by subtracting the baseline, which was averaged 5 s before stimulation, from the capacitance value after stimulation. Fast capacitance jump and total capacitance were, respectively, measured as the value immediately after stimulation or as the peak value (before the capacitance started to decline again owing to endocytosis). Ca²⁺ currents were isolated pharmacologically with a bath solution containing (in mM): 105 NaCl, 20 TEA‐Cl, 2.5 KCl, 1 MgCl₂, 2 CaCl₂, 25 NaHCO₃, 1.25 NaH₂PO₄, 25 glucose, 0.4 ascorbic acid, 3 myo‐inositol, 2 Na‐pyruvate, 0.001 tetrodotoxin (TTX), 300–310 mOsm, and pH 7.4 when bubbled with 95% O₂ and 5% CO₂. The standard pipette solution contained the following (in mM): 125 Cs‐gluconate, 20 CsCl, 4 MgATP, 10 Na₂‐phosphocreatine, 0.3 NaGTP, 10 HEPES, 0.05 BAPTA, 310–320 mOsm, and pH 7.2 adjusted with CsOH. Patch pipettes with resistance of 3–4 MΩ were used, and series resistances, typically < 20 MΩ, were compensated by 70–90%.

Presynaptic BK‐current recordings

Presynaptic BK‐currents were studied using whole‐cell patch‐clamp technique similar as Ca²⁺ current recording as described above. BK‐currents were recorded used a pipette solution that contained the following (in mM): 97.5 potassium gluconate, 32.5 KCl, 10 HEPES, 1 MgCl₂, 12 sodium phosphocreatine, 2 ATP‐Mg, 0.5 GTP‐Na, and 0.2 EGTA (305 mOsm, pH 7.3). The standard ACSF solution with added 1 μM TTX and 2.5 mM 4‐aminopyridine was used. The BK‐currents were pharmacologically blocked by 200 nM iberiotoxin (IBTX) and isolated by subtracting the currents recorded from the same terminal before and after 5–6 min of IBTX perfusion.

Immunohistochemistry

Mice of postnatal day 12–14 were anesthetized and perfused with 1× PBS for 5 min followed by 2–4% paraformaldehyde (PFA) for 5 min. The brains were carefully extracted and post‐fixed in 4% PFA for 2 h, followed by immersion in 20–30% sucrose for 48 h for complete cryo‐protection. Transverse brain sections at 20–30 μm were cut at −20°C using a cryostat (CM3050S, Leica). The slices containing the MNTB nucleus were pretreated in 0.5% Triton X‐100 and 5% goat serum in PBS for 1 h at room temperature and incubated overnight at 4°C with primary antibodies in blocking solution (0.1% Triton X‐100 and 5% goat serum in PBS). The slices were washed with PBS and incubated with fluorescence‐conjugated secondary antibodies for 2 h at room temperature. After wash, the slices were mounted with DAPI fluoromount (SouthernBiotech). Primary antibodies against Syt2 (rabbit, polyclonal, 1:1,000, A320, RRID: AB_2650431), VGluT1 (guinea pig, polyclonal, 1:1,000, Millipore, Cat#: AB5905; RRID: AB_2301751), VGAT (rabbit, polyclonal, 1:500, Millipore, Cat#: AB5062P), and GluA1 (rabbit, monoclonal, 1:500, Millipore, Cat#: 04‐855), Bassoon (mouse, monoclonal, 1:200, SySy, Cat#: 141011), CaV2.1 (rabbit, polyclonal, 1:200, SySy, Cat#: 152203) were used. Secondary antibodies were Alexa Fluor conjugates (1:500; Invitrogen). Images were acquired using Nikon A1RSi confocal microscope with a 60× oil‐immersion objective (1.45 numerical aperture) and analyzed in Nikon Analysis software.

Immunoblot

Brain slices containing MNTB nucleus were cut as described above for electrophysiology. The MNTB areas were isolated mechanically with forceps, and tissues from two pups were pooled and homogenized in 50 μl of 0.32 M ice‐cold sucrose buffer containing (in mM): 1 HEPES, 1 MgCl₂, 1 EDTA, 1 NaHCO₃, and 0.1 PMSF at pH 7.4, in the presence of a complete set of protease inhibitors (Complete; Roche Diagnostics, Basel, Switzerland). Triton X‐100 at a final concentration of 1% was added to the homogenization solution. Tissues were incubated for 30 min on ice and centrifuged to 20,000 g for 10 min. Supernatants were collected, and protein concentration was quantified using the Bradford Assay (Bio‐Rad Protein Assay 500‐0006, München, Germany). 5–10 mg proteins was separated and analyzed by immunoblot.

5–10 mg proteins was separated by SDS–PAGE using 10 wells, 4–20% mini protean TGX precast gels (Bio‐Rad). Proteins were then transferred onto nitrocellulose membranes for 10 min at 2.5 V using the Trans‐Blot Turbo Transfer System (Bio‐Rad). Membranes were blocked in Tris‐buffered saline (5% no‐fat milk powder, 0.1% Tween20) for 1 h at room temperature. Primary antibodies were diluted in the same buffer and incubated overnight at 4°C. The following antibodies were used at 1:1,000 dilution: RBP2 (4193, RRID:AB_2617050), RIM1‐2 (U1565, RRID:AB_2617054), Synaptotagmin 2 (znp‐1, ZFIN, RRID:AB_10013783), vGlut1 (AB5905, Millipore, RRID:AB_2301751), CASK (75‐000, Neuromab, RRID:AB_2068730), PSD‐95 (75‐028, Neuromab, RRID:AB_2307331), and Tubulin (T2200, Sigma, RRID:AB_262133). Combinations of the following IRDye secondary antibodies were used (1:10,000 dilution): IRDye 800CW donkey anti‐mouse (926‐32212), IRDye 680LT donkey anti‐mouse (926‐68022), IRDye 800CW donkey anti‐rabbit (926‐32213), and IRDye 680LT donkey anti‐rabbit (926‐68023), from LI‐COR. Pseudo colors were then applied to the signals. Detection of the signal was obtained by Odyssey CLx imaging systems (LI‐COR). Quantification was performed with Image Studio 5.2 free software.

Quantifications and statistical analyses

Electrophysiological data were analyzed in Igor Pro (Wavemetrics), except miniature synaptic currents were analyzed in MiniAnalysis (Synaptosoft). For clarity, all stimulus artifacts were blanked and not showed in the figures. All data were shown as means ± SEM. The numbers of analyzed cells from at least two mice per group were shown in the graph as indicated in the figures. Student's t‐test was used for two‐group comparisons, except Kolmogorov–Smirnov test was used for cumulative distributions. Statistical significance was defined and indicated in the figures and figure legends as follows: *P < 0.05; **P < 0.01; and ***P < 0.001.

Author contributions

FL and TCS designed and analyzed all experiments and wrote the paper; FL, AS, and MJ performed the experiments.

Conflict of interest

The authors declare that they have no conflict of interest.

Supporting information

Expanded View Figures PDF

Review Process File

The EMBO Journal (2020) 39: e103208