Presynaptic voltage-gated calcium channels in the auditory brainstem

Sound information encoding within the initial synapses in the auditory brainstem requires reliable and precise synaptic transmission in response to rapid and large fluctuations in action potential (AP) firing rates. The magnitude and location of Ca2+ entry through voltage-gated Ca2+ channels (CaV) in the presynaptic terminal are key determinants in triggering AP-mediated release. In the mammalian central nervous system (CNS), the CaV2.1 subtype is the critical subtype for CNS function, since it is the most efficient CaV2 subtype in triggering AP-mediated synaptic vesicle (SV) release. Auditory brainstem synapses utilize CaV2.1 to sustain fast and repetitive SV release to encode sound information. Therefore, understanding the presynaptic mechanisms that control CaV2.1 localization, organization and biophysical properties are integral to understanding auditory processing. Here, we review our current knowledge about the control of presynaptic CaV2 abundance and organization in the auditory brainstem and impact on the regulation of auditory processing.


Introduction
Fundamental to hearing is the ability to accurately identify and interpret sound information over a wide range of time scales down to submilliseconds (Pickles, 2013;Schnupp et al., 2011). An inability to do so results in problems with speech perception and sound localization in the aging population, central auditory processing disorder (CAPD) and auditory deficits in autism spectrum disorders, and leads to cognitive impairments (Atcherson et al., 2015;Fitzgibbons and Gordon-Salant, 1996;Griffiths et al., 2020;Humes et al., 2012;Roque et al., 2019;Smith et al., 2019). The ability to encode the accurate identification and perception of sound requires synaptic transmission in the first few synaptic connections in the lower auditory brainstem to drive and sustain precise firing of action potentials (APs) over rapid and large fluctuations of AP firing rates (Borst and Soria van Hoeve, 2012;Friauf et al., 2015;Grothe et al., 2010). The magnitude and location of Ca 2+ in the presynaptic terminal regulates the probability and kinetics of synaptic vesicle (SV) release, which controls information flow between neurons (Dittman and Ryan, 2019;Stanley, 2016) (Fig. 1). In the mammalian central nervous system (CNS) synapses, the Ca V 2 family of voltage-gated Calcium channels (VGCCs), Ca V 2.1, Ca V 2.2, and Ca V 2.3, regulate synaptic transmission, with Ca V 2.1 being the dominant subtype regulating CNS function (Nanou and Catterall, 2018;Simms and Zamponi, 2014).
In the majority of CNS synapses, Ca V 2.1 is the most effective at triggering AP-mediated release (Iwasaki et al., 2000;Takahashi and Momiyama, 1993;Wheeler et al., 1994), since it is most abundant and in closer proximity to SV release sites than Ca V 2.2 and Ca V 2.3 channels (Eggermann et al., 2012;Wu et al., 1998;Wu et al., 1999). During neuronal circuit maturation, synaptic transmission becomes increasingly Ca V 2.1-dependent due to a selective presynaptic reduction or loss of Ca V 2.2 and Ca V 2.3 in many synapses (Iwasaki et al., 2000;Iwasaki and Takahashi, 1998;Scholz and Miller, 1995). In addition, during development of neurons that signal with rapid and temporally precise APs in the mature circuit, SV release kinetics become faster as distances between Ca V 2.1 and SV release sites become shorter (Baur et al., 2015;Chen et al., 2015;Eggermann et al., 2012;Fedchyshyn and Wang, 2005;Matsui and Jahr, 2003;Nakamura et al., 2015;Stanley, 2016;Voinova et al., 2015).
While the hair cells in the cochlea are essential for sound detection, the auditory brainstem is composed of specific neurons in distinct nuclei that are responsible for the computations required for processing the temporal features of sound and the localization of sound sources (Pickles, 2013) (Fig. 2). Unlike hair cell synapses which contain specialized active zone (AZ) structures and utilize Ca V 1 channels to encode sensory information in an analog fashion (Pangrsic et al., 2018;Wichmann and Moser, 2015), auditory brainstem synapses have conventional AZ ultrastructures, use presynaptic Ca V 2 channels with few exceptions, and utilize AP-evoked synaptic transmission to encode auditory information (Grothe et al., 2010;Leao, 2019). Therefore, elucidating the mechanisms that control presynaptic Ca V 2 channel abundance and organization in auditory brainstem synapses is fundamental to understanding the key underpinnings of the initial stages of auditory information processing and how defects in these processes lead to auditory processing disorders. Here we will summarize our current knowledge of presynaptic Ca V 2 channel abundance and organization in auditory brainstem synapses and review the mechanisms implied to regulate presynaptic Ca V 2 subtype levels and organization. Finally, we highlight future directions regarding Ca V 2 channel research that will be of importance to understanding auditory signaling.

Voltage-gated Ca 2þ channel complex composition
VGCCs are the Ca 2+ entry sites in the presynaptic AZ that are essential for triggering SV release in response to APs (Nanou and Catterall, 2018). In the CNS, VGCCs are multi-subunit complexes consisting of an α 1 , β, and α 2 δ, and sometimes the γ subunit (Simms and Zamponi, 2014) (Fig. 3A). The α 1 subunit is the critical pore forming subunit that determines VGCC kinetics, voltage dependence, pharmacology, and conductance. In the mammalian CNS, the α 1 subunit is encoded by ten different genes which give rise to three different VGCC families, Ca V 1 (Ltype), Ca V 2 (P/Q, N, R type), and Ca V 3 (T-type). The Ca V 2 family is composed of three members, Ca V 2.1 (P/Q), Ca V 2.2 (N), and Ca V 2.3 (R) which are the dominant channels in the mammalian CNS that mediate AP-evoked SV release. The pore-forming Ca V 2 α 1 subunit contains multiple cytoplasmic interaction sites that bind to various AZ proteins (Fig. 3B). These interactions are implicated in regulating presynaptic Ca V 2 abundance and control of SV release kinetics within the presynaptic AZ (Catterall, 1999(Catterall, , 2011Felix, 2005;Lubbert et al., 2019;Lubbert et al., 2017;Muller et al., 2010;Simms and Zamponi, 2014). Ca V 2 subtype abundance in the presynaptic membrane are different from those found in the neuron soma membrane (Doughty et al., 1998;Fisher and Bourque, 1995;Miki et al., 2013), therefore the mechanism(s) that control presynaptic and somatic Ca V 2 abundance must be distinct.
Three cytoplasmic regions in the Ca V 2 α 1 subunit are implicated as critical for regulating presynaptic Ca V 2 subtype abundance and the position of Ca V 2 channels relative to SV release sites (defined as coupling) (Simms and Zamponi, 2014), a dominant parameter controlling SV release kinetics (Neher, 2015). These three motifs are: 1) loop I-II region, which contains the primary Ca V β subunit binding site (AID) 2) the loop II-III region, which contains the synprint motif (the interaction site with the SNARE proteins) and 3) the C-terminal region, which contains many motifs that mediate binding to AZ proteins (Simms and Zamponi, 2014) and is frequently mutated in Ca V 2.1 channelopathies (Pietrobon, 2010) (Fig. 3B). Since these regions are not highly conserved between the different Ca V 2 isoforms (Zamponi, 2016;Zamponi et al., 2015) and are subject to alternative splicing (Lipscombe et al., 2013), they are speculated to regulate the differences between the abundance and organization of presynaptic Ca V 2 subtypes through protein-protein interactions (Catterall, 2011;Felix, 2005;Lubbert et al., 2019;Simms and Zamponi, 2014).
The Ca V β, α 2 δ, γ subunit are auxiliary subunits that are encoded by distinct gene families (Nanou and Catterall, 2018). In mammals, four distinct genes, CACNB1, CACNB2, CACNB3, CACNB4, encode the Ca V β1, Ca V β2, Ca V β3, Ca V β4 subunits, are heavily spliced, which results in many different splice variants (Buraei and Yang, 2013;Rima et al., 2016). The Ca V β subunits are localized intracellularly and differentially impact the voltage dependence of activation and inactivation (Buraei and Yang, 2013) . Furthermore, they are critical for trafficking Ca V 2 channels to the plasma membrane where they are involved in regulating surface level expression. Studies analyzing Ca V β subunits affinity for the Ca V 2 α 1 subunit indicate that Ca V β4 has the highest affinity compared to the other isoforms with order of affinities from strongest to weakest being De Waard et al., 1995). Finally, splicing generates many different Ca V β isoforms which differentially modulates regulation of these pathways (Buraei and Yang, 2013;Rima et al., 2016).
The α 2 δ subunits are encoded by four different genes, CACN2D1, CACN2D2, CACN2D3, CACN2D4 and are extensively spliced (Ablinger et al., 2020;Dolphin, 2013). They are extracellular membrane proteins, that are implicated in regulating Ca V 2 trafficking, surface expression and voltage dependent activation of Ca V 2 channels (Dolphin, 2013). They have important roles in synapse formation, structure and function that are independent of regulating Ca V 2 trafficking and membrane levels (Ablinger et al., 2020;Geisler et al., 2015). Eight different genes in the CACNG family, CACNG1-8, encode the γ subunits (Chen et al., 2007). Although the γ subunit is expressed in the brain, evidence for a role in Ca V 2 trafficking in the CNS is lacking.
Ca V 2.1 channels are the dominant Ca V 2 subtype in the majority of presynaptic terminals and the most efficient in triggering AP-mediated release, while Ca V 2.3 is the least efficient (Dietrich et al., 2003;Simms and Zamponi, 2014;Takahashi and Momiyama, 1993;Wheeler et al., 1994;Wu et al., 1998). Ca V 2 subtypes are differentially modulated by Gprotein coupled receptors, metabotropic receptors, and the ability to undergo Ca 2+ current facilitation, which impacts synaptic transmission Fig. 1. Voltage-Gated Calcium channel number and position relative to SVs impacts AP-evoked release. A) An increase in VGCC number results in increased Ca 2+ entry, SV release and the magnitude of excitatory postsynaptic current (EPSC). B) Ca V 2 proximity to SVs, defined as coupling, underpins the effectiveness of Ca V 2 subtypes in eliciting AP-evoked release and SV release kinetics. and plasticity (Huang and Zamponi, 2017;Nanou and Catterall, 2018). Global knock out (KO) of Ca V 2.1 in mice has severe effects on animal viability which results in lethality by ~3 weeks (Jun et al., 1999). Global KO of Ca V 2.2 results in partial lethality and global KO of Ca V 2.3 in mice has no impact on viability (Saegusa et al., 2001;Saegusa et al., 2000). Finally, Ca V 2.1 is the dominant Ca V 2 isoform associated with Ca V 2 channelopathies that manifest in neurological disorders in the CNS (Carbone and Mori, 2020;Heyes et al., 2015;Pietrobon, 2010). Taken together these data, indicate the importance of Ca V 2.1 channels relative to Ca V 2.2 and Ca V 2.3 in the CNS.

Presynaptic Ca V 2 channels in the auditory brainstem
In the auditory brainstem, the Ca V 2 channel family are the key VGCCs for driving AP-evoked synaptic transmission. Currently, studies examining presynaptic Ca V 2 subtype composition have been carried only in a few synapses in the lower auditory brainstem: the endbulbs of Held-spherical bushy cell (SBC) synapse, the calyx of Held-medial nucleus of the trapezoid body (MNTB) synapse, the MNTB-lateral superior olive (LSO) synapse, the MNTB-medial superior olive (MSO) synapse, and the SBC-MSO synapse (Alamilla and Gillespie, 2013;Barnes-Davies et al., 2001;Inchauspe et al., 2007;Iwasaki and Takahashi, 2001;Lin et al., 2011;Oleskevich and Walmsley, 2002;Zhuang et al., 2020) (Fig. 2).

Endbulbs of Held
The endbulbs of Held which arise from the ascending branches of the auditory nerve fibers are large presynaptic terminals which contact the SBCs in the cochlear nucleus and are required to drive temporally precise AP signaling (Brawer and Morest, 1975;Wichmann and Moser, 2015). Direct presynaptic whole-cell patch clamp recordings of Ca 2+ currents at prehearing endbulbs of Held revealed that Ca V 2.1 currents are the dominant Ca V 2 current (upwards of ~90%) (Lin et al., 2011). Four days after hearing onset, although the Ca V 2.2 channels can elicit AP-evoked synaptic transmission, Ca V 2.1 channel blocker ω-agatoxin IVA reduced AP-evoked release by ~90% (Oleskevich and Walmsley, 2002). Analysis of adult animals found that ω-agatoxin IVA almost completely blocked endbulb AP-evoked release (~95%) on average, however a small amount of sensitivity to the Ca V 2.2 subtype blocker, ω-conotoxin GVIA still persisted (Zhuang et al., 2020). Based on these results the endbulbs of Held still contain a small population of Ca V 2.2 channels relative to Ca V 2.1 channels in the mature stage. Non-stationary fluctuation analysis of presynaptic Ca 2+ currents estimate that the entire endbulb of Held contains on average ~6400 Ca V 2 channels (Lin et al., 2011). Morphological studies based on ultrathin section transmission electron microscopy indicate there are on average 155 active zones per endbulb of Held (Nicol and Walmsley, 2002). Based on these approximations, each presynaptic active zone in the endbulb contains ~40 Ca V 2 channels, of upwards to on average 36 Ca V 2.1 channels.
The role of sound-evoked activity in modulating Cav2 subtype levels at the endbulbs of Held has been explored. Deafness mutant mice which are congenitally deaf from birth exhibited no change in their Ca V 2 subtype contribution to AP-evoked release at the endbulb (Oleskevich and Walmsley, 2002). However, experiments that manipulated soundevoked activity levels though noise rearing or occlusion of the ear canal after the onset of hearing found that noise rearing does not impact Ca V 2 channel abundance while ear canal occlusion does (Zhuang et al., 2020). In particular, ear canal occlusion resulted in a significant reduction of Ca V 2.1 currents with a concomitant increase in Ca V 2.2 currents. Therefore, this indicates the mechanisms that regulate presynaptic Ca V 2 channel abundance at the endbulb of Held are impacted by the manner of sound-evoked activity manipulation and the developmental stage it is performed.

The calyx of Held
The calyx of Held, a giant axosomatic glutamatergic presynaptic terminal in the medial nucleus of the trapezoid body neuron (MNTB) arises from the globular bushy cell axon (Held, 1893;Morest, 1968). The calyx of Held is the sole input that relays afferent AP spike patterns from the cochlear nucleus to the MNTB, which in turn provides rapid and precise glycinergic inhibition to key mono-and binaural cell groups (Joris and Trussell, 2018). Direct whole-cell presynaptic patch clamp recordings at the calyx of Held and analysis of AP-evoked synaptic transmission at the calyx/MNTB synapse demonstrated that before hearing onset, the calyx of Held is a mixed Ca V 2 subtype terminal, with Ca V 2.1 being the dominant subtype Doughty et al., 1998;Iwasaki et al., 2000;Iwasaki and Takahashi, 1998;Wu et al., 1999). In contrast, after hearing onset there is a shift to Ca V 2.1 exclusivity at the calyx of Held (Doughty et al., 1998;Iwasaki and Takahashi, 1998).
Since the globular bushy neuron soma contains Ca V 1, Ca V 2, and Ca V 3 channels and their levels are distinct of the developmental changes at the presynaptic terminal (Doughty et al., 1998), this indicates that change in presynaptic Cav2 abundance are regulated by local mechanism(s) and not global mechanism(s) regulating Ca V 2 channel abundance. Analysis of transcription of Ca V 2 subtype levels at the pre-and post-hearing stage at the calyx of Held indicate no appreciable change in the different Ca V 2 α 1 subunit mRNA levels (Korber et al., 2014). Furthermore, viral vector-mediated overexpression of the Ca V 2.1 α 1 subunit before the onset of hearing at the wild-type mouse calyx of Held increased Ca V 2.1 currents and channel numbers and at the prehearing state there was a concomitant decrease in Ca V 2.2 current amplitude (Lubbert et al., 2019). However, overexpression of Ca V 2.2 α 1 subunit before and after hearing did not impact Ca V 2.1 levels (Lubbert et al., 2019). Ablation of Ca V 2.1 results in an increase in presynaptic Ca V 2.2 currents but not Ca V 2.3 currents before and after the onset of hearing (Inchauspe et al., 2004;Ishikawa et al., 2005;Lubbert et al., 2017). However, the total presynaptic Ca 2+ current in the Cav2.1 null calyx is reduced compared to wild-type levels (Inchauspe et al., 2004;Ishikawa et al., 2005;Lubbert et al., 2017).
Estimates using non-stationary noise analysis at the prehearing mouse calyx of Held estimate a total of ~20,000 Ca V 2 channels (Lin et al., 2011). Given that the average mouse calyx has ~400 active zones, each calyx of Held active zone will contain ~50 Ca V 2 channels (Lin et al., 2011). This estimate is in line with Sodium-Dodecyl-Sulfate Freeze Fracture Immunolabeling (SDS-FFRIL) EM measurements, which found that putative AZs contain ~20-80 Ca V 2.1 channels at both the pre-and post-hearing calyx (Lubbert et al., 2019;Nakamura et al., 2015). Although the total presynaptic Ca 2+ current remains similar at the preand post-hearing calyx of Held, presynaptic Ca V 2.1 channel numbers in the AZ increase during development (Lubbert et al., 2019;Nakamura et al., 2015). Thus, although unitary presynaptic Ca V 2.1 currents may increase during development, the dominant mechanism responsible for maintaining the total presynaptic Ca 2+ currents is the increase in presynaptic Ca V 2.1 channel numbers with the concomitant loss of presynaptic Ca V 2.2 and Ca V 2.3 currents. Deaf mice (Ca V 1.3 − /− ) do not have a difference in the presynaptic Ca V 2 current amplitudes at the calyx (Erazo-Fischer et al., 2007), suggesting that sound evoked activity is not responsible for setting presynaptic Ca V 2 levels. However, it is unknown if the manner of sound-evoked activity manipulation and the developmental stage impacts presynaptic Ca V 2 subtype levels at the calyx. Recent findings indicate that individual calyx of Held AZs are not saturated with Ca V 2.1 channels, as Ca V 2.1 α 1 subunit overexpression at P1 or P14 resulted in an increased number of Ca V 2.1 channels at the AZ, resulting in increased synaptic strength (Lubbert et al., 2019). However, the mechanisms that regulate Ca V 2 subtype levels are unknown.

Lateral superior olive
Synapses in the lateral superior olive (LSO) are critical for detecting interaural intensity differences (Grothe et al., 2010;Yin et al., 2019). Studies from mice two-three days after hearing onset demonstrate that synaptic transmission at GABA/glycinergic projections from the MNTB to the LSO are ~75-90% dependent on Ca V 2.1, while the remaining contribution is equally split between Ca V 1 and Ca V 2.2 channels (Giugovaz-Tropper et al., 2011). Developmental profiles indicate that similar to the endbulbs and the calyx of Held, total Ca V 2 currents are constant but there is a dramatic reduction in Ca V 1 and Ca V 2.2 currents during the progression to the onset of hearing (Alamilla and Gillespie, 2013;Giugovaz-Tropper et al., 2011). Glutamatergic transmission in the immature LSO also follows a similar developmental pattern of presynaptic Ca V 2 channel subtype contribution prior to these synapses switching to purely glycinergic transmission (Alamilla and Gillespie, 2013). This indicates that the presynaptic Ca V 2 channel composition is independent of the neurotransmitter utilized. Similar to the calyx, ablation of Ca V 2.1 in the MNTB/LSO synapse leads to Ca V 2.2 dominance (on average ~80% Ca V 2.2) with a slight increase in Ca V 1 currents (Giugovaz-Tropper et al., 2011).

Medial superior olive
The medial superior olive (MSO) is critical for encoding interaural timing differences (Grothe et al., 2010;Yin et al., 2019). Analysis of presynaptic Ca V 2 current contribution to AP-evoked release at the glycinergic MNTB-MSO synapse revealed that both Ca V 2.1 and Ca V 2.2 channels contribute to AP-evoked release (Barnes-Davies et al., 2001). However, Ca V 2.1 channels are the dominant subtype. Analysis of the contralateral glutamatergic spherical bushy cell-MSO synapse found that both Ca V 2.1 and Ca V 2.2 equally contributed to AP-evoked release at hearing onset (Barnes-Davies et al., 2001). However, if Ca V 2.2 channel contribution at this synapse is decreased in the mature auditory brainstem is unknown. Based on these synapses in auditory brainstem nuclei that control timing and intensity processing, the mechanisms that increase Ca V 2.1 channel abundance during development are likely shared. and auxiliary subunits. B) The three Ca V 2 isoforms share 54% consensus conservation in the cytoplasmic C-terminus (red circle). Key protein binding sites shared by Ca V 2.1 and Ca V 2.2 in C-terminus: RIM1/2, MINT1, RBP, and CASK. The domain II-III linker, which contains the synprint region (green circle) has 65% consensus. Syntaxin IA and SNAP 25 interact with synprint region of Ca V 2.1 and Ca V 2.2. The domain I-II linker which contains the Ca V β subunit binding site (purple circle) has 91% consensus. Positions are listed as amino acid sequences from Mus musculus CACNA1A (accession no: NP_031604.3).

Presynaptic Ca V 2 channel organization in the auditory brainstem synapses
In presynaptic terminals a key determinant that regulates synaptic transmission and plasticity is the organization of VGCCs with respect to SVs (Nusser, 2018;Stanley, 2016;Wang and Augustine, 2014). Presynaptic terminals contain SVs either tightly or loosely coupled to Ca V 2, which directly correlates to the SV release modes utilized by synapses (Baur et al., 2015;Eggermann et al., 2012;Fedchyshyn and Wang, 2005;Neher, 1998;Neher and Sakaba, 2008) (Fig. 4). Microdomain synapses have fast and slow SV release rates and broad AP waveforms, which produce both synchronous and asynchronous release in response to APs. Since the AP waveform is relatively broad, Ca 2+ can diffuse over distances of 50-100 nm in response to an AP. These synapses utilize many loosely coupled VGCCs (~100 nm between the SVs and VGCC cluster) to trigger SV release and synaptic transmission is sensitive to EGTA, a slow binding Ca 2+ chelator. Microdomain release is supported by all three Ca V 2 subtypes and used by many synapses in neuronal circuits that require presynaptic plasticity to encode dynamic changes in response to repetitive stimulation (Hefft and Jonas, 2005;Ohana and Sakmann, 1998;Rozov et al., 2001;Vyleta and Jonas, 2014). Nanodomain synapses are Ca V 2.1 exclusive and have very fast SV release rates and a very narrow AP waveform, producing only synchronous release which is EGTA insensitive. Since the AP waveform is narrow, Ca 2+ diffusion is relatively constrained to under 50 nm in response to APs. Nanodomain release is utilized by synapses in neuronal circuits that require rapid and temporally precise APs to encode information (Baur et al., 2015;Bucurenciu et al., 2008;Chen et al., 2015;Fedchyshyn and Wang, 2005;Hefft and Jonas, 2005;Matsui and Jahr, 2003). However, despite being Ca V 2.1 dominant, presynapses in immature neuronal circuits that encode temporal fidelity transition from microdomain to nanodomain release (Baur et al., 2015;Fedchyshyn and Wang, 2005). Therefore, these data suggest that the Ca V 2.1 channel is not the sole instructive signal for nanodomain release, and it is possible that mechanisms that facilitate reorganization of AZ proteins within the AZ are key.
Our current knowledge about calcium channel organization and control of SV release kinetics in the auditory brainstem is largely based on the calyx of Held. Prior to the onset of hearing, AP-evoked synaptic transmission at the calyx is EGTA sensitive (Borst and Sakmann, 1996;Fedchyshyn and Wang, 2005). However, at the mature calyx, AP-evoked synaptic transmission is relatively EGTA insensitive and the Ca 2+ cooperativity of release decreases during development (Fedchyshyn and Wang, 2005). Therefore, it is well-accepted that the developmental transition to nanodomain release at the calyx of Held is critical for temporal coding of auditory signals (Joris and Trussell, 2018). Based on analysis of the AP-evoked release sensitivity to EGTA, it has been hypothesized that SVs in the prehearing calyx are ~60 nm away from Ca V 2 clusters (Borst and Sakmann, 1996;Meinrenken et al., 2002;Wadel et al., 2007) and ~20 nm from Ca V 2 clusters in the mature calyx . However, these original studies were not based on morphological data that directly measured Ca V 2 channel organization in the presynaptic AZ. Therefore, modeling simulations on Ca V 2 channel arrangements found many potential solutions to how the Ca V 2 channels could be arranged to replicate AP-evoked release (Keller et al., 2015;Meinrenken et al., 2002;Wang et al., 2009). A major breakthrough came from studies using SDS-FFRIL detection of Ca V 2.1 channels on the calyx release face, which revealed that Ca V 2.1 channels are organized as clusters . Although SDS-FFRIL is unable to report SV positions at the AZ (Fujimoto, 1995(Fujimoto, , 1997, by combining the experimentally derived channel organization with 3D reaction-diffusion modeling simulations of AP-evoked SV release it was determined that a perimeter of SVs around clusters of Ca V 2.1 channels was able to reproduce SV release in response to APs . Unlike other models based only on AZ morphology and simulated SV and Ca V 2.1 organization (Keller et al., 2015), the perimeter model was able to reproduce the developmental changes in AP-evoked SV release kinetics and Ca 2+ cooperativity of release. Subsequent studies using the perimeter model found that during early development SVs in the readily releasable pool (RRP), the pool of SVs that are released by APs, are located on average ~50-60 nm from Ca V 2 channels, while in the mature calyx of Held the pool of SVs that can be release by APs are ~5-25 nm away from Ca V 2.1 channels (Chen et al., 2015). Therefore, based on these studies the developmental shift in the tightening of SV coupling to Ca V 2.1 channels is the dominant parameter that underpins the fast Fig. 4. Release modes utilized by synapses to code information. Microdomain synapses (VGCC-SV distance ~50-100 nm) have broad APs, utilize many loosely coupled VGCCs, and have slower SV release rates which result in synchronous and asynchronous release. Nanodomain synapses (VGCC-SV distance <25 nm) have narrow APs, use a few tightly coupled VGCCs, and have fast SV release rates which result in synchronous release. temporal coding by the calyx of Held.
Although mature calyx of Held utilizes SVs tightly coupled to Cav2.1 channels, there is heterogeneity in the morphology and synaptic transmission properties (Grande and Wang, 2011). Since morphologically simple calyces of Held have distinct release properties compared to morphologically complex calyces of Held, it has been proposed this morphological functional continuum at the calyx of Held/MNTB synapse potentially expands the coding capacity of sound information (Grande and Wang, 2011). What underpins this morphologicalfunctional continuum? Individual AZs in the calyces of Held contain a broad range of Cav2.1 channel numbers (Dong et al., 2018;Lubbert et al., 2019;Nakamura et al., 2015). In addition, individual AZs contain a mixture of SVs that are tightly and loosely coupled to Cav2.1 channels and it is possible that there are individual AZs that only contain loosely coupled SVs in the mature calyx of Held (Chen et al., 2015). Therefore, differences in Cav2.1 channel abundance and coupling distance may contribute to the morphological-functional continuum. Recent work, using GFP-tagged Cav2.1 knock in mice in combination with 3D reconstruction, Ca 2+ imaging and simulations using the perimeter model provided evidence that differences in coupling distances between SVs and Cav2.1 channels are key morphological determinants driving synaptic heterogeneity at the mature calyx of Held (Fekete et al., 2019). Currently, studies on the molecular determinants that establish heterogeneity are in their early stages. Septin5 is the only known molecule in the field that has been demonstrated to regulate microdomain to nanodomain transition at the calyx of Held (Yang et al., 2010). Furthermore, at mature calyx of Held synapses, Septin5 perturbation had different impacts on synaptic transmission at calyces with either complex or simple structures (Fekete et al., 2019). Therefore, this clearly demonstrates the importance of Septin5 in regulating synaptic heterogeneity (Fekete et al., 2019). In addition, the Munc13 family is implicated in regulating SV to Cav2.1 channel coupling at the calyx of Held (Chen et al., 2013). Although the calyx of Held contains all three Munc13 isoforms, the Munc13-1 isoform is the only isoform highly localized in the AZ (Chen et al., 2013). Most importantly, absence of Munc13-2 and Munc13-3 isoforms results in faster RRP release kinetics at the calyx of Held (Chen et al., 2013). Therefore, it is possible that heterogeneity in Septin5 levels and Munc13 isoform levels at individual AZs in the mature calyx of Held are key determinants of the morphological-functional continuum that is proposed to expand the coding capacity for a broad spectrum of sound information. It should be noted, the role of the morphological-functional continuum at the calyx of Held in encoding sound information remains to be tested.
Currently, the organization of the Ca V 2.2 and Ca V 2.3 channels in the presynaptic AZ of the prehearing calyx is unknown. Unfortunately, the current commercially available antibodies for these channels that worked in SDS-FFRIL in hippocampal synapses do not work at the calyx (unpublished observations). However, analysis of EGTA sensitivity of Ca V 2.2 and Ca V 2.3 channel-mediated AP-evoked release indicates that Ca V 2.2 and Ca V 2.3 channels are located further away from SVs than Ca V 2.1 channels at the calyx of Held (Wu et al., 1998;Wu et al., 1999). Subsequent analysis of AP-evoked SV release from Ca V 2.1 KO animals found that release was more sensitive to EGTA, which supports the idea that Ca V 2.2 channels are located more distal to SVs (Inchauspe et al., 2007;Inchauspe et al., 2004). However, since total Ca 2+ currents are decreased in the Cav2.1 KO animal it is difficult to draw a definite conclusion if this the change is solely due to changes in distance or changes in presynaptic Ca 2+ currents. Previous studies have shown that EGTA sensitivity of AP evoked release in the presence or absence of conotoxin is similar (Wu et al., 1999). Therefore, it has been proposed that Ca V 2.1 and Ca V 2.2 channels are located at similar distances respective to SVs in the presynaptic AZ (Wu et al., 1999). Studies at the endbulbs of Held that modulated sound-evoked activity using noise rearing or ear canal occlusion found no impact on the Ca 2+ cooperativity of release (Zhuang et al., 2020). Since the Ca V 2.2 contribution to APevoked release is increased with ear canal occlusion (Zhuang et al., 2020), this suggests that Ca V 2.2 channels in the endbulb of Held are located at similar distances from SVs as Ca V 2.1 channels. However, how other forms of deafness or modulation of sound-evoked activity impact Ca 2+ cooperativity and SV coupling at other auditory brainstem synapses are unknown.
Analysis of SV release kinetics at both the endbulbs of Held (Lin et al., 2011) and the endbulbs of ventral nucleus of the lateral lemniscus (VNLL) demonstrated that they have similar SV release kinetics as the calyx of Held (Berger et al., 2014). Since these terminals are proposed to be scaled down versions of the calyx of Held, it is possible that they have similar Ca V 2 channel arrangements. However, the arrangement of presynaptic Ca V 2 channels in synapses that encode interaural intensity differences are unknown. Recent work based on modeling studies from granule cell presynaptic terminals in the cerebellum, which do not use temporal coding, suggest these smaller synapses use an alternate Ca V 2 channel arrangement than at the calyx of Held and stellate cell presynaptic terminals in the cerebellum (Rebola et al., 2019). Therefore, it is possible this alternate arrangement may be similar in intensity encoding synapses or other synapses in the auditory brainstem that are not required for temporal coding (Keller et al., 2015;Rebola et al., 2019).

Molecular mechanisms controlling presynaptic Ca V 2 abundance and organization
Although many proteins interact with the Ca V 2 channel complex (Muller et al., 2010), the molecular mechanisms that control presynaptic Ca V 2 subtype levels independent of somatic and dendritic levels in their native environment are unsolved. While studies using cell culture have provided insight into possible mechanisms controlling presynaptic Ca V 2 levels, they have yielded conflicting results, leaving uncertain whether these mechanisms are essential in vivo (Acuna et al., 2015;Atasoy et al., 2007;Butz et al., 1998;Cao et al., 2004;Cao and Tsien, 2010;Davydova et al., 2014;Hibino et al., 2002;Ho et al., 2006;Kaeser et al., 2011;Maximov et al., 1999;Meyer et al., 2019;Spafford et al., 2003a;Spafford et al., 2003b;Szabo et al., 2006;Wong et al., 2013;Wong et al., 2014;Wong and Stanley, 2010). Studies at the calyx of Held have demonstrated that mechanisms regulating Ca V 2 coupling are independent of those regulating Ca V 2 abundance (Acuna et al., 2015;Lubbert et al., 2017). However, the mechanisms that establish Ca V 2.1 exclusivity or dominance and Ca V 2 organization in auditory brainstem presynaptic terminals are largely unknown. Due to the central role of the Ca V 2 α 1 subunit in regulating synaptic transmission through a multitude of interactions, we primarily highlight the Ca V 2 α 1 subunit motifs and protein interactions implied to regulate presynaptic Ca V 2 levels and abundance and their potential role in the auditory brainstem.

Ca V β subunits
The Ca V β subunits are critical for regulating Ca V 2 α 1 subunit trafficking, surface level expression and gating (Buraei and Yang, 2013;Rima et al., 2016). They bind to the Ca V 2 α 1 subunit with high affinity at the α-interaction domain (AID) of the I-II loop Pragnell et al., 1994). Although a secondary β interaction exists in the Ca V 2 α 1 subunit (Walker et al., 1998), loss of that region had no impact on Ca V 2.1 levels at the prehearing calyx of Held (Lubbert et al., 2017). In the auditory brainstem, four Ca V β isoforms are variably expressed (Allen Mouse Brain Atlas) (Lein et al., 2007) and microarray analysis revealed that globular bushy cells express all four Ca V β isoforms (Korber et al., 2014). However, Ca V β isoform localization within the calyx of Held and other auditory brainstem presynaptic terminals is currently unknown. Although global Ca V β knock out mouse lines exists (Buraei and Yang, 2013), a complete characterization of Ca V β isoforms using these Ca V β KO mouse lines in the auditory brainstem has not been carried out. Ca V β4 and Ca V β2a are located within the cultured hippocampal neuron presynaptic terminals (Xie et al., 2007) and Ca V 2.1 channels have higher specificity for binding to Ca V β4 than Ca V 2.2 and Ca V 2.3 channels (Muller et al., 2010). Therefore, presynaptic Ca V β4 isoform levels may play a role in regulating presynaptic Cav2.1 levels and organization. In addition, the Ca V β4 is alternatively spliced into two functionally distinct isoforms Ca V β4a and Ca V β4b, with the Ca V β4a splice variant having distinct protein-protein interactions independent of channel gating (Vendel et al., 2006a;Vendel et al., 2006b). Since these two isoforms are differentially expressed in distinct neuronal cell types, alternative splicing of the Ca V β4 subunits may be a critical determinant controlling presynaptic Ca V 2.1 levels and coupling.
Ca V β subunits directly interact with RIM1/2s (Kiyonaka et al., 2007) and CAST/ELKS proteins (Kiyonaka et al., 2012). Ablation of these proteins at the calyx of Held has shown that they are key regulators of presynaptic Ca V 2 current amplitudes and Ca V 2 channel numbers (Dong et al., 2018;Han et al., 2015;Han et al., 2011;Radulovic et al., 2020). Since ablation of both CAST/ELKS and RIMs result in similar phenotypes, it is possible that direct interactions between Ca V β, RIMs and CAST/ELKS serve as the key building blocks of the central organizing scaffold regulating presynaptic Ca V 2 channel abundance and organization (Dong et al., 2018;Kiyonaka et al., 2007;Lubbert et al., 2017) (Fig. 5).

α 2 δ subunits
α 2 δ subunits are auxiliary components of the Ca V channel complex which are critical for Ca V 2 channel trafficking, surface level expression, biophysical properties and synapse development (Dolphin, 2016(Dolphin, , 2018. They are proposed to bind the extracellular the L5 loops of the Cav2 α 1 subunit between repeats I to III (Wu et al., 2015). In cultured hippocampal neurons, α 2 δ1 overexpression led to increased Ca V 2 channel abundance (Hoppa et al., 2012;Schneider et al., 2015) and increases in AZ size (Schneider et al., 2015). In addition, since α 2 δ1 overexpression increased the efficacy of AP-evoked released despite a reduction in presynaptic Ca 2+ influx, this implicates that α 2 δ subunits are controlling Ca V 2 coupling to SVs through unknown interaction partners (Hoppa et al., 2012).
However, use of these animals to characterize the role of α 2 δ subunits in the auditory brainstem has been minimal. Global α 2 δ3 knock out results in a reduction in Ca V 2.1 channel abundance in the soma of spiral ganglion neurons, a reduction in the size of smaller auditory nerve fiber synapses in the cochlear nucleus, and dramatic impacts on auditory signaling (Pirone et al., 2014). Studies analyzing Ca V 2 currents in cultured spiral ganglion neurons from 3 week old α 2 δ3 null mice showed a severe reduction in somatic Ca V 2.1 and Ca V 2.3 currents whereas somatic Ca V 2.2, Ca V 1 or Ca V 3 currents remained unaltered (Stephani et al., 2019). However, in P5 α 2 δ3 null spiral ganglion neurons, somatic Ca V 2.1 and Ca V 2.2 currents were not reduced, but Ca V 1 currents were reduced (Stephani et al., 2019). In contrast, the cortex of α 2 δ3 null mice showed increased Ca V 2.2 and Ca V 2.3 channel abundance but with no change in Ca V 2.1 channel abundance (Landmann et al., 2018). Regardless, neither of these studies compared Ca V 2.1 abundance at both the presynapse and soma. Moreover, the phenotypes seen with global knock out mouse lines may be due to homeostatic compensation in the CNS. Therefore, whether α 2 δ3 regulation of Ca V 2 levels is due to global mechanisms or mechanisms specific to the presynaptic terminal remain to be tested.

Synprint domain
The synprint region, which binds to SNARE proteins, is located in the Ca V 2 α 1 subunit domain II-III linker and is implicated in Ca V 2 trafficking, plasma membrane abundance and voltage-dependent activation (Nanou and Catterall, 2018). This region is highly variable between different Ca V 2 α 1 subunits. In addition, the Ca V 2.3 channel does not contain a syntaxin 1A binding site (Jurkat-Rott and Lehmann-Horn, 2004;Rajapaksha et al., 2008;Simms and Zamponi, 2014). Studies that swapped synprint regions between the Ca V 1 and Ca V 2 channels demonstrated that the Ca V 2 synprint region was necessary for trafficking to the presynaptic terminal and AP-evoked release in cultured superior cervical ganglion (SCG) neurons (Mochida et al., 2003a;Mochida et al., 2003b). Furthermore, these studies demonstrated no difference in APevoked release with Ca V 2.1 chimeras containing the Ca V 2.2 synprint region. This suggests that in cultured SCG neurons, the synprint regions in Ca V 2.1 and Ca V 2.2 are not critical for differences in presynaptic localization between Ca V 2.1 and Ca V 2.2 channels (Mochida et al., 2003a;Mochida et al., 2003b). Further analysis of AP-evoked cholinergic release at SCG neurons revealed that the syntaxin 1A binding site in the synprint domain was critical for Ca V 2.1 and Ca V 2.2 channel abundance (Mochida et al., 2003a;Mochida et al., 2003b). In contrast, syntaxin 1A binding sites are dispensable to support AP-evoked release in hippocampal neurons (Szabo et al., 2006). Furthermore, MNTB-LSO synapses contain presynaptic Ca V 1 channels at the same level as Ca V 2.2 channels in Cav2.1 KO mice (Giugovaz-Tropper et al., 2011). Thus, how the mechanisms on the importance of the synprint region regulation of Ca V 2 levels and organization in CNS neurons is unclear.
The synprint domain is extensively spliced in the syntaxin 1A binding domain (Rajapaksha et al., 2008). Therefore, differences in syntaxin 1A binding affinity could change Ca V 2.1 presynaptic abundance and coupling. Ca V 2.1 α 1 subunit isoforms lacking a large portion of the synprint region containing the syntaxin 1A binding sites have reduced plasma membrane incorporation in neuroendocrine cells (Kamp et al., 2005;Rajapaksha et al., 2008). However, whether these synprint Ca V 2 splice variants are present in auditory brainstem neurons is currently unknown. In addition, the Ca V 2.2 and Ca V 2.3 α 1 subunit mRNA contains an alternatively spliced exon 18a in the synprint region that is not found in Ca V 2.1 (Lipscombe et al., 2013). Therefore, it is possible that exon18a splicing may control Ca V 2 subtype trafficking and incorporation into the plasma membrane at the presynaptic terminal. The Ca V 2.2 exon18a variant levels are upregulated and found in distinct neuron types that predominately express Ca V 2.2 channels in the presynaptic terminal (Gray et al., 2007;Jurkat-Rott and Lehmann-Horn, 2004).

Presynaptic active zone molecules
Many presynaptic active zone proteins are implicated to control Ca V 2 abundance and coupling through interactions in the Ca V 2 α 1 subunit (Chen et al., 2018;Sudhof, 2012). RIM1/2 proteins are large presynaptic AZ molecules that can directly bind the C-terminal DDXC motif in the Ca V 2 α 1 subunit . They are important for regulating presynaptic Cav2 channel density and abundance in both invertebrate and vertebrate synapses, as deletion of RIMs results in significant reductions in Ca V 2 current amplitudes (Han et al., 2011;Kaeser et al., 2011;Liu et al., 2011). Furthermore, deletion of RIM1/2 at the prehearing calyx of Held results in a slight slowing of SV release rates (Han et al., 2011). MINT1 proteins are part of the tripartite complex with CASK and Velis (Butz et al., 1998;Maximov et al., 1999). MINT1 proteins directly interact with the Ca V 2 α 1 subunit through the same DDXC motif as RIMS1/2, which was necessary and sufficient for trafficking of Ca V 2.2 to the presynaptic terminals in immature primary culture hippocampal neurons (Maximov and Bezprozvanny, 2002). CASK proteins potentially regulate Ca V 2.2 abundance as part of the tripartite complex with MINT1. RIM binding proteins (RBPs) bind to the Ca V 2 family via a conserved PQTPLTPRP motif and may regulate Ca V 2.2 coupling to increase neurotransmitter release in neuroendocrine cells (Hibino et al., 2002). Further, RBPs are proposed to specifically regulate Ca V 2.1 abundance through a direct interaction with Bassoon (Davydova et al., 2014).
Despite the importance of these molecules, the mechanisms of action of these proteins is ambiguous. RIMs, RBPs, and MINTs direct interaction with Ca V 2.1 and 2.2 α 1 subunits is proposed to be critical for setting Cav2 levels and controlling coupling (Han et al., 2011;Hibino et al., 2002;Kaeser et al., 2011). However, deletion of these direct interaction motifs in the Ca V 2.1 C-terminus had no impact on presynaptic Ca V 2.1 levels or SV release kinetics at the prehearing calyx (Lubbert et al., 2017). In addition, a unique region in the Cav2.1 α 1 subunit was identified that play a role in the regulation of SV release kinetics and the readily releasable pool size (Lubbert et al., 2017). Overexpression of Ca V 2.1 α 1 splice variants lacking the RIM, RBP, or MINT1 binding sites in a Ca V 2.1 KO background was able to rescue the Ca V 2.1 contribution to AP-evoked release (Cao et al., 2004;Cao and Tsien, 2010). Furthermore, biochemical assays have failed to detect a direct interaction between the Cav2.1α 1 subunit and RIM, RBP, and MINT1 (Wong et al., 2013;Wong et al., 2014;Wong and Stanley, 2010). In addition, ablation of MINT1 or CASK proteins in mice has no effect on AP-evoked release (Atasoy et al., 2007;Ho et al., 2006). However, deletion of RBPs leads to slower SV release rates with no change in Ca V 2 abundance (Acuna et al., 2015), indicating that RBPs do not regulate presynaptic Ca V 2 levels but may play a role in SV coupling. Therefore, these direct interactions with either RIM1/2 , MINT1, CASK (Maximov and Bezprozvanny, 2002;Maximov et al., 1999) and RBP (Davydova et al., 2014;Hibino et al., 2002) proteins are not essential for controlling presynaptic Ca V 2.1 levels and SV release in microdomain release mode synapse. However, it is possible these interactions regulate SV release in nanodomain release mode synapse.
Bassoon is the only AZ protein proposed to specifically control presynaptic Ca V 2.1 channel abundance (Davydova et al., 2014). However, rescue experiments with Ca V 2.1 channels lacking the bassoon binding interaction domain at the calyx of Held demonstrated that Ca V 2.1 channel abundance were similar to wild-type Ca V 2.1 channel abundance (Lubbert et al., 2017). In addition, knockdown of bassoon or dual knockout of bassoon and piccolo at the post hearing calyx, which is Ca V 2.1 exclusive, has no effect on basal AP-evoked release (Parthier et al., 2017). Furthermore, analysis of Ca V 2.1 and Ca V 2.2 mEOS-tagged protein levels in cultured hippocampal neurons appear to be equivalent at bassoon positive puncta (Schneider et al., 2015). Therefore based on these data, a role for bassoon in specifically controlling presynaptic Ca V 2.1 channel abundance is unlikely, although it is possible that bassoon may play a role in setting presynaptic Cav2.1 abundance in other synapses.
The CAST/ELKS proteins are an evolutionarily conserved core presynaptic AZ protein family found in invertebrate and vertebrate synapses (Deken et al., 2005;Ohtsuka et al., 2002;Wagh et al., 2006;Wang et al., 2002) that are key regulators of presynaptic calcium channel abundance (Dong et al., 2018;Kittel et al., 2006;Radulovic et al., 2020). In the auditory brainstem, studies on CAST/ELKS function have been restricted to the calyx of Held (Dong et al., 2018;Radulovic et al., 2020). Deletion of both CAST/ELKS at the prehearing and posthearing calyx of Held resulted in a reduction in presynaptic Ca V 2 channel currents and Ca V 2 numbers. The mechanism by which CAST/ELKS regulates Ca V 2.1 levels and biophysical properties is unknown. CAST/ELKS directly bind with high affinity to the Ca V β subunit (Kiyonaka et al., 2012) and to the RIM PDZ domain through the DDXC motif in the C-terminus of CAST/ ELKS (Lu et al., 2005). In addition, CAST/ELKS has a weaker direct interaction with the Ca V 2.1α1 subunit in the loop II-III domain (Kiyonaka et al., 2012). Although the data favors a model in which CAST/ ELKS is a key regulator (Fig. 5), the relevance of these interactions in vivo remain to be tested.

Conclusions
In this review we highlighted our current knowledge on presynaptic voltage-gated calcium channel abundance and organization in auditory brainstem synapses and the potential regulatory mechanisms. These studies have shown that presynaptic Ca V 2.1 channels are the dominant subtype regulating synaptic transmission and that auditory brainstem synapses increase their reliance of presynaptic Cav2.1 during neuronal circuit maturation. Interestingly, it appears that Cav2 subtype levels can be impacted by the manner of sound-evoked activity manipulation and the developmental stage it is performed. Based on these observations, this raises a multitude of questions. What are the molecular mechanisms that regulate presynaptic Ca V 2 abundance and organization in the auditory brainstem? How does manipulation of auditory activity impact presynaptic Ca V 2 channel abundance and organization? Why does the type of sound-evoked activity manipulation and developmental time period differentially impact presynaptic Ca V 2 subtype levels? In addition, MNTB-LSO contain presynaptic Ca V 1 channels. Is their presynaptic presence unique to MNTB-LSO synapses or are they found in other presynaptic terminals in the auditory brainstem? What is the role that presynaptic Cav2.2, Cav2.3 and Cav1 channels play during development of the auditory brainstem? Do changes in presynaptic Cav2 channels and organization in the auditory brainstem synapses contribute to auditory processing disorders?
Currently studies on presynaptic channels have been limited to a few synapses in the auditory brainstem. Given that presynaptic VGCCs can have vastly different biophysical, pharmacological, and modulatory properties, it is critical to further investigate other presynaptic terminals in the auditory brainstem. The molecular diversity within presynaptic terminals enables a wide range of synaptic properties (Nusser, 2018). Therefore, there will likely be a multitude of presynaptic mechanisms such as trafficking, insertion, retention and stability that regulate presynaptic VGCC levels and organization in the AZ (Lubbert et al., 2019;Lubbert et al., 2017). New genetic models and molecular tools will be needed that will allow for cell specific and temporal specific manipulation of presynaptic VGCCs. Fundamental and important work remains to be done to elucidate our understanding on how presynaptic VGCCs regulate the localization of sound sources, processing of the temporal features of sound, and the contributions of defects in these processes to auditory processing disorders.

Declaration of competing interest
We have no competing interests to declare. I am signing on behalf of myself and Dr. Priyadharishini Veeraraghavan.