Excess neurotransmitters in the synapse that goes back into the presynaptic neuron

The organization of the presynaptic terminal is one important element for optimization of secretion and recycling

The presynaptic terminal is compact so that vesicles may rapidly move from one organelle compartment to another. Unlike endocrine secretion, where membrane fusion occurs randomly on a relatively large membrane area, synaptic vesicles are secreted only at active zones after docking has taken place. The process of docking means that vesicles are primed for membrane fusion in response to highly localized changes in Ca2+. The ability to dock at an active zone is unique to synaptic vesicles. The peptidergic, dense core vesicles also found in most presynaptic terminals are released through unspecialized membrane regions, and the kinetics of peptide secretion is similar to release of hormones from endocrine cells. The composition of synaptic vesicles, the active zone and many of the components needed for docking have been identified (Murthy & De Camilli, 2003; Sudhof, 2004) (see Table 7-2 for a list of proteins and putative functions).

Table 7-2. A Glossary of Proteins in the Synapse

One characteristic of regulated exocytosis is the ability to store secretory vesicles in a reserve pool for utilization upon stimulation. In the presynaptic terminal, this principle is expanded to define multiple pools of synaptic vesicles: a readily releasable pool, a recycled synaptic vesicle pool and a larger reserve pool. This reserve pool assures that neurotransmitter-containing vesicles are available for release in response to even the highest physiological demands. Neurons can fire so many times per minute because synaptic vesicles from the readily releasable pool at a given synapse undergo exocytosis in response to a single action potential. Those vesicles have been primed by docking at the active zone and are therefore ready for exocytosis upon arrival of an action potential. However, for the synapse to respond rapidly and repeatedly under heavy physiological demand, these exocytosed vesicles must be rapidly replaced. This is accomplished first from the recycled pool of vesicles and, as the demand increases, from the reserve pool. To be recycled, synaptic vesicles must be reloaded quickly after they release their contents. The sequence of events that is triggered by neurotransmitter exocytosis is known as the synaptic vesicle cycle (Murthy & De Camilli, 2003; Sudhof, 2004) (Fig. 7-8). In neurons, an action potential triggers a highly localized rise of synaptic intracellular Ca2+ through opening of voltage-gated Ca2+ channels closely apposed to the active zones. These Ca2+ ions bind proteins that serve as Ca2+ sensors, among them specific synaptotagmins, which in turn trigger the vesicular membrane fusion with the plasma membrane for docked synaptic vesicles (Pang & Sudhof, 2010).

The Synaptic Vesicle Cycle

At the presynaptic terminals, SVs are not of ‘single use.’ They are regenerated at the terminals independent from protein synthesis in the cell body. The SV cycle can be outlined, with the uptake of neurotransmitter into SVs as a first step (Figure 2(b), step 1). Neurotransmitters in the central nervous system (CNS) are synthesized locally in the cytoplasm of the presynaptic terminals and are actively transported into SVs. Away from the plasma membrane, the majority of neurotransmitter-filled SVs are either diffusively floating in the cytoplasm or tethered with cytoskeleton components such as actin and spectrin (step 2). For exocytosis, SVs must come into physical contact with the plasma membrane (docking; step 3). SVs do not evenly dock to the whole area of the presynaptic plasma membrane but, rather, to a restricted area called the active zone. There, large cytomatrix proteins, such as bassoon, piccolo, and munc13, form huge protein complexes that appear as electron-dense structures on electron micrographs. Docked SVs are then transformed into fusion-competent SVs via a process called priming (step 4). As soon as an electrical stimulus arrives at the terminal, voltage-dependent calcium channels at the active zone open, resulting in a rapid and local increase in Ca2+ concentration. Ca2+ ions trigger the fusion of the SV membrane with the plasma membrane in less than 100 μs. Since other exocytotic reactions (i.e., hormone secretion from endocrine cells) take much longer (seconds to minutes), there must be unique factors present only in neurons to perform such a rapid membrane fusion reaction.

After exocytosis, SV components that are incorporated into the plasma membrane are retrieved to form a new SV by endocytosis. There are at least two kinetically distinct modes of endocytosis. The time constants of the fast and slow phase are approximately 1 and 10 s, respectively. Whereas the molecular machinery for the fast phase is not well understood, the slow phase is mediated by the formation of clathrin-coated pits. In both modes, GTP hydrolysis by the GTPase protein dynamin is indispensable for the fission of the invaginated membranes of newly formed SVs. Although various endocytosis-related proteins, such as AP-2, endophilin, amphiphysin, and synaptojanin, have been identified and implicated in controlling endocytosis, their precise roles are a matter of intense research.

The reformed SVs then either recycle back and are refilled with neurotransmitters (step 1) or pass through the early endosomal intermediates before recycling back to step 1. An alternative pathway has been proposed that is similar to the exocytosis of secretory granules; that is, SVs do not fully collapse with the plasma membrane upon fusion but instead form a narrow and transient fusion pore which does not allow a complete discharge of neurotransmitter content. As soon as the pore closes, the half-empty SV can either be immediately engaged in another round of exocytosis or go back to step 1. The existence of such a rapid recycling mode in the CNS, termed the ‘kiss-and-run’ mechanism, is under debate.

The SVs clustering at the presynaptic terminals can be divided into two functional pools. The first pool contains a small fraction of SVs (5–10% of the total SVs at the presynaptic terminals) that can be released rapidly by a brief high-frequency train of action potentials or by stimulation with hypertonic solution. This pool is thought to be release-ready and is therefore referred as to the readily releasable pool (RRP). The second pool, the reserve pool (RP), represents a vesicle fraction that does not immediate participate in exocytosis. Instead of participating in exocytosis, the RP vesicles replenish the RRP pool after the RRP vesicles undergo exocytosis. Both the amount of the RRP and the rate of replenishment of RRP with RP are critical parameters to determine the availability of vesicles for exocytosis, thereby affecting the characteristics of short- and long-term plasticity of a given neuron. Classically, the RRP was related to a fraction of morphologically docked vesicles at the plasma membrane and the RP was thought to be spatially distant from the plasma membrane. However, one study suggested that the RRP vesicles do not necessarily correlate with the morphologically docked vesicles but are randomly distributed in the SV cluster at the terminal, indicating that there are no correlations between the anatomical locations of SVs and their functional fusion competence. Mechanisms underlying how the mobility and the fate of an individual SV can be molecularly defined are uncertain.

G-Protein-Coupled Receptor Mediated Modulation at the Presynaptic Terminal

Regulation of neurotransmitter release at the presynaptic terminal plays an important part in the plasticity of the nervous system [44]. Various neurotransmitters modulate release from presynaptic terminals, and many of these interactions involve the activation of a G-protein-coupled receptor (GPCR) [45]. Modulation of exocytotic release by GPCRs is an important mechanism by which neurons are able to respond and adapt to changes in secretory requirements. Some GPCRs may couple to more than one G protein, while others show a great deal of specificity. Gβγ binding to Gα involves widespread contacts at two distinct interfaces. Following activation by a GPCR, the heterotrimeric G protein dissociates into an activated Gα-GTP subunit and a free Gβγ subunit [46] Active Gα-GTP and free Gβγ may then activate many different signaling pathways [47].

Uncertainty over the mechanisms by which G proteins alter neurotransmitter release in part reflects the variety of G-protein effector targets and the difficulties in gaining experimental access to these small structures. Thus, most molecular studies of the detailed mechanisms come from either transfection of the relevant proteins into cultured cell lines and Xenopus oocytes, or from electrophysiological measurements from neuronal cell bodies.

Presynaptic Calcium Dynamics

Calcium influx into the presynaptic terminal regulates the release of synaptic vesicles [3]. Multiple buffering mechanisms tightly control intracellular calcium concentration. Dedicated molecules such as synaptotagmin act as calcium sensors, detecting changes in calcium concentration and triggering the release of neurotransmitter by fusion of vesicles filled with neurotransmitter molecules with the membrane (exocytosis). Synapses typically consist of several release sites, and upon calcium influx in response to a presynaptic action potential, there is a certain probability for a vesicle to be released at each release site. This probability is termed release probability, p(release), often abbreviated as p. Every vesicle contains a constant amount of neurotransmitter molecules, denoted as q (abbreviation for quantum of neurotransmitter inside the vesicle). Therefore only a small number of different, discrete quantities of neurotransmitter can be released at a synapse. If a synapse has N release sites, the amount of neurotransmitter that can be released ranges from zero to Nq.

We need to revisit probability theory to understand how we can use measurements of the mean and the variance of synaptic strength to determine p and N. First, let us consider an example. If there are N = 5 release sites and we know that m = 3 sites actually did release a vesicle for a given presynaptic action potential, there are several different sets of three release sites that may have released neurotransmitter (for example: sites 1, 2, and 3; or sites 2, 3, and 4). The number of possible sets is given by how many different sets of three sites can be drawn from the total set of five. Mathematically, this number is called N choose m (combination), denoted as:

(3.1)(Nm)=N!(N−m)!m! =N(N−1)…(N−m+1)m(m− 1)…1

where n factorial is defined as N! = N(N − 1)(N − 2)…1. In our example, we find that 5 choose 3 = (5∗4∗3)/(3∗2∗1) = 10. In other words, there are 10 different outcomes that all result in three vesicles released. To determine the probability with which any three vesicles are released, we need to determine with what joint probability release of three vesicles occurs and then multiply this probability with 10. If we assume that the release probabilities at individual release sites are independent from each other, we can multiply the individual probabilities of release to get the joint probability. Therefore, the joint probability with what release of three vesicles occurs is given by p∗p∗p∗(1 − p)∗(1 − p), where the first three factors stand for the sites where release occurred and the last two for the sites where no release occurred. Finally, the probability of any three vesicles being released is given by 10p3(1 − p)2. In general, the joint probability of m out of N sites releasing neurotransmitter is:

(3.2)Pr(m)=pm(1−p)N−m

Since we do not care which of the N release sites provided neurotransmitter (as long as we know how many did), we can combine these equations to find the binomial distribution that describes the probability of any m out of N sites being active:

(3.3)p(m)=(Nm)pm(1−p)N−m

The mean μ and variance σ2 of this binomial distribution are:

(3.4)μ=Np

(3.5)σ2=Np(1−p)

Therefore we can use measurements of postsynaptic events and determine their mean and variance to compute N and p (we use “hats” in our notation to indicate that those values are estimates):

(3.6)pˆ=1−σ2μ

(3.7)Nˆ=μpˆ

Synapses also release neurotransmitter vesicles (typically one) in the absence of presynaptic action potentials. These so-called miniature postsynaptic potentials (often referred to as minis or mPSPs) can be isolated by application of tetrodotoxin, a drug that blocks voltage-gated sodium channels and therefore action potential firing. The functional role of minis is still undetermined. Recording minis provides another way to estimate the value of q since only one vesicle is released at a time.

Presynaptic Effects of Kainate Receptors

Activation of kainate receptors on presynaptic terminals results in the facilitation or suppression of neurotransmitter release of a wide variety of transmitter systems. Infusion of 5- or 20-μM kainate into the prefrontal cortex increases dopamine concentrations to about 150% or 500% above baseline levels. However, infusion of either 20- or 100-μM AMPA produces an increase in dopamine up to about 250% or 1500% above baseline, respectively. This effect is blocked by CNQX, an antagonist to both AMPA and kainate receptors, demonstrating that the facilitation of dopamine release is mediated by both AMPA and kainate receptors.

Interneurons and pyramidal neurons within the BLA express kainate receptors that are localized to both somatodendritic and axonal sites. In particular, the GluK1 subunit expression is higher relative to other subtypes within this brain region. Application of ATPA, a selective agonist of GluK1 subunit–containing kainate receptors, to BLA pyramidal neurons differentially alters the failure rate of evoked inhibitory postsynaptic currents (eIPSCs) in a dose-dependent fashion. The 300-nM APTA decreases the eIPSC failure rate, whereas both 1-and 10-μM ATPA increases the eIPSC failure rate. This indicates that synaptic release of GABA is facilitated by the low-level activation of presynaptic kainate receptors on interneurons but is suppressed at higher levels of activation.

The presynaptic role of kainate receptors on glutamatergic neurons has also been studied. Evoked excitatory postsynaptic currents (eEPSCs) recorded from pyramidal cells are dose-dependently blocked by the application of UBP302, a selective antagonist to GluK1 subunit–containing kainate receptors. Application of ATPA decreases the failure rate of eEPSCs and increases the frequency of miniature excitatory postsynaptic currents (mEPSCs), demonstrating that kainate receptors are also acting presynaptically in the BLA to facilitate glutamate release. Taken together, these studies provide an explanation of how kainate receptor stimulation can produce seizures from the BLA, a known seizurogenic structure. As higher concentrations of agonist increase the failure rate of eIPSCs and decrease the failure rate of eEPSCs, this may suggest a shift in the BLA network’s ratio of excitation to inhibition when kainate receptors are overstimulated.

Summary

Signals exchanged at developing synapses ensure that differentiated presynaptic terminals are aligned precisely with a highly specialized postsynaptic membrane. Analysis of mice lacking innervation demonstrates that a coarse pattern of postsynaptic differentiation is present prior to innervation and that neuronal signals, including Agrin, both maintain and induce postsynaptic differentiation at sites of nerve–muscle contact (Fig. 17.2). We are gaining a better understanding of the pathways that act downstream from MuSK to control postsynaptic differentiation (Fig. 17.5), but we remain poorly informed about the signals and pathways that control presynaptic differentiation.

Neurotransmission in the nervous system is initiated at presynaptic terminals by fusion of synaptic vesicles with the plasma membrane and subsequent exocytic release of chemical transmitters. There are multiple methods to detect neurotransmitter release from nerve terminals. Most commonly employed methods monitor actions of released chemical substances on postsynaptic receptors or artificial substrates such as carbon fibers. These methods are closest to the physiological setting because they have a rapid time resolution and they measure the action of the endogenous neurotransmitters rather than the signals emitted by exogenous probes. However, postsynaptic receptors only indirectly report neurotransmitter release in a form modified by the properties of receptors, which are often nonlinear detectors of released substances. In the past decade, in addition to electrophysiological and biochemical methods, several fluorescence imaging modalities have been introduced which report synaptic vesicle fusion, endocytosis, and recycling. These methods either take advantage of styryl dyes that can be loaded into recycling vesicles or exogenous expression of synaptic vesicle proteins tagged with a pH-sensitive green fluorescent protein variant at regions facing the vesicle lumen. This article provides an overview of these methods, with emphasis on their relative strengths and weaknesses, and it discusses the types of information one can obtain from them.

Receptor clustering and postsynaptic differentiation at the NMJ

The aggregation of neurotransmitter receptors apposed to the presynaptic terminal is a principal feature of synaptogenesis. Since this process has been most thoroughly described at the vertebrate neuromuscular junction (NMJ), we will discuss this well-studied synapse, before turning to postsynaptic differentiation in the central nervous system.

Is receptor clustering a cell-autonomous process, or is it induced by the presynaptic terminal? At first inspection, the postsynaptic site appears to be produced in an autonomous fashion. Acetylcholine receptors (AChRs) form small clusters on the muscle cell membrane before the motor axons arrive (Fischbach and Cohen, 1973). Structures that resemble postsynaptic densities, but with no apparent presynaptic element, have also been found in the developing olfactory bulb and visual cortex during early development (Hinds and Hinds, 1976; Bahr and Wolff, 1985). In fact, most neurotransmitter receptors are expressed before innervation occurs, and they are often found in clusters, similar in appearance to a postsynaptic site (Figure 8.9A).

At the time of innervation, the muscle cell membrane still displays an immature distribution of AChRs. This was originally demonstrated by recording the response from rat muscle cells in vivo as ACh was applied at different places along the myofiber surface (Diamond and Miledi, 1962). Early in development, ACh application at each site evokes a similar shift in membrane potential. As the muscle is innervated, the ACh-evoked response becomes much larger at the site of innervation, and the response at extrasynaptic regions declines. Methods were subsequently developed to visualize the distribution of AChRs by labeling them with radioactive α-bungarotoxin (α-Btx), a high-affinity peptide from the venom of the Taiwanese cobra (Bevan and Steinbach, 1977; Burden, 1977a). Consistent with the electrophysiological measures, α-Btx labeling is broadly distributed at first and then becomes highly localized to the synapse. The process of clustering leads to a dramatic disparity in receptor concentration: there are >10,000 AChRs/μm2 at the synaptic region but <10/μm2 in extrasynaptic regions (Fertuck and Salpeter, 1976; Burden, 1977a; Salpeter and Harris, 1983).

While these observations implied that the motor axons induce receptor clustering, higher AChR concentrations occur autonomously at the center of developing skeletal muscle fibers (Yang et al., 2001). This was demonstrated in a mouse mutant where the diaphragm muscle remains uninnervated during embryonic development, yet AChRs nonetheless become concentrated in the central muscle (Figure 8.20). In fact, this initial localization of AChRs is caused by a postsynaptic clustering system that will ultimately become responsive to a motor nerve-released signal for stabilization (see next section, below, for discussion of Agrin-MuSK signaling).

Although the localization of AChRs is, at first, independent of motor axons, the nerve terminal does exert a potent clustering influence when it arrives. This was first suggested by observing fluorescently-labeled AChRs during the period of innervation (Anderson and Cohen, 1977; Cohen et al., 1979). Although small AChR clusters are present prior to innervation, they do not serve as preferential sites of innervation. Rather, the growing neurites induce the rapid accumulation of AChRs as they extended across the muscle (Figure 8.21). Furthermore, the ability to induce AChR clusters is specific to spinal cord neurons, which presumably include motor neurons (Cohen and Weldon, 1980; Kidokoro et al., 1980). However, the AChRs do not have to be activated in order to cluster. When myocytes and spinal cord are cultured in the presence of an AChR antagonist, clustering occurs normally at the site of neurite contact in the absence of cholinergic transmission (Cohen, 1972; Anderson and Cohen, 1977).

27.2.1 Presynaptic Molecular Composition

A variety of proteins that are associated with the presynaptic terminal have been identified, and they can be grouped into several protein classes by function (Dresbach et al., 2001; Schoch and Gundelfinger, 2006; Ziv and Garner, 2004). The first are proteins associated with SVs that are critical for their localization and calcium-dependent fusion (Chua et al., 2010; Sudhof, 2004; Takamori et al., 2006), including the vesicle-associated membrane protein 2 (VAMP2), synaptotagmin, and synaptic vesicle glycoprotein 2 (SV2). The second class comprises a variety of several different scaffolding proteins, including Piccolo, Bassoon, Ca2 +/calmodulin-dependent serine protein kinase (CASK), Munc18-interacting protein 1 (MINT1), synapse-associated protein 97 (SAP97), Rab3-interating molecule (RIM), and liprin-α, which interact with each other to form a multimolecular structure at the active zone. These scaffolding proteins tether SVs in close proximity to both endocytic and exocytotic machineries and link them to a variety of signaling pathways (Altrock et al., 2003; Ohtsuka et al., 2002; Schoch and Gundelfinger, 2006; Takao-Rikitsu et al., 2004). Finally, there are numerous cytoskeletal proteins, such as actin, myosin, tubulin, and β-catenin, that provide the structural framework for the presynaptic terminal, regulating the distribution of the many active zone proteins (Doussau and Augustine, 2000; Halpain, 2003). Together, through physical interaction, these classes of proteins create a highly organized network of precisely aligned signaling complexes that enable the Ca2 +-dependent release and retrieval of SVs for synaptic transmission and plasticity.