Introduction

A synapse (gr. syn “together” and haptein “grasping”) refers to the contact between two neurons at which, by the release of chemical messengers, information is transmitted. It is typically divided into a presynapse, a synaptic cleft and a postsynapse. A huge number of synapses (10¹³) in humans occur between neurons of the central nervous system where they provide the foundation of plastic processes such as learning, memory and dementia. Beta cells in the pancreas and chromaffin cells of the adrenal gland use similar mechanisms to release the hormones insulin and adrenalin, respectively, which then circulate via the blood stream and affect many target cells in the entire body. Both natural killer cells of the innate immune system and T-lymphocytes of the adaptive immune system build, in the hunt for pathogens, highly specific contact zones with antigen-presenting cells. These contact zones are the site of targeted release of cytotoxic substances which kills these cells and have been named immunological synapses. Genetic studies in patients with immune-deficiencies indicate that release of cytotoxic substances at immunological synapses is astonishingly similar to release of messengers at neuronal synapses and neuroendocrine cells.

Though referred to as synapses, there are important differences between immunological synapses and neuronal synapses. There is little similarity between target cell recognition by T-lymphocytes, discussed below, and the recognition process at neuronal synapses, a fascinating area (Akins and Biederer, 2006) that is beyond the scope of this article. We will concentrate on presynaptic mechanisms since immunological synapses have no requirement for generating postsynaptic structures, a requirement for true synapses. The adrenal chromaffin cell, which we use as a model for regulated exocytosis, makes no contact with its target cells.

Both neurons and chromaffin cells generate large reserves of vesicles and presynaptic machinery which organizes transfer of vesicles to the releasable pools via docking and priming. T-lymphocytes lack a reserve pool per se and do not maintain a releasable pool which is discussed below. T-lymphocytes do, however, express components of the fusion machinery and do carry out carry out docking and priming steps.

In the last decade, mutation of the proteins Syntaxin11, Munc13-4, Munc18-2a and und Rab27a have been shown to result in loss of immunological synapse function, which lead to lethal immune diseases (Janka, 2012). These results indicate fundamental similarities between release from lymphocytes, neuronal and neuroendocrine cells. In contrast to synapses in the CNS and neuroendocrine tissue, immunological synapses are built within 30 minutes and are rapidly disassembled so that new targets can be acquired. Thus, in addition to study of synapse function, study of the assembly and disassembly of these synapses with high resolution methods is possible. The availability of human material from donor blood allows use of lymphocytes and study of the immunological synapse provides results which may be extended to neuronal synapses. In the case of the presynaptic organization of release and recycling of the release machinery, this appears to be the case. In this overview we briefly summarize the processes of assembly, release and disassembly at immunological synapses and will discuss similarities and differences of these processes to those in chromaffin cells.

Assembling the Synapse

Neuronal synapses form in response to intercellular interactions of adhesion molecules (Akins and Biederer, 2006). Both neuroligins and SynCAM1 have been reported to serve as markers for postsynaptic targets. When neurons come in contact with either marker, even when it is present in non-neuronal cells, presynaptic specialization is induced in the neuron. Adhesion molecules cluster at these contact points and attract other associated molecules such as those involved in active zone formation. The molecules involved in the active zone and in the release machinery interact with binding sites located on the cytosolic tails of adhesion molecules and with each other. In both neurons and chromaffin cells, large multi-domain proteins related to Bassoon and Piccolo act as scaffolds and thus organize the structures which transfer vesicles from the large reserve pool to the docking and priming apparatus (Gundelfinger et al., 2015). These organizing molecules are delivered to the presynaptic compartment as passengers on transport vesicles containing active zone components such as piccolo, bassoon, syntaxin, RIM, Munc-18, ELKS2/CAST, SNAP-25 and n-cadherin. These vesicles travel along microtubles (Bury and Sabo, 2016). These active zone molecules and their transport vesicles are generated in the soma, as are synaptic vesicles and LDCVs. In the case of neurons, the transport in axons may be over long distances. In neuroendocrine cells such as chromaffin cells the transport distance is much shorter, but the organizing principles are similar.

The active zone proteins not only organize docking and priming, but also attract the voltage-dependent calcium channels required for stimulated release. It has been difficult to assign specific functions to individual molecules. Studies examining single knockout or knockdown have produced surprisingly modest phenotypes (Südhof, 2012). It may be that the multiple interactions of these multi-domain proteins produce a functional redundancy, in particular in tissue that has already gone through development.

Chromaffin cells of the adrenal medulla control important physiological parameters such as blood pressure and heartbeat via secretion the catecholamines, adrenalin and noradrenaline. The catecholamines are stored in large dense core vesicles (LDCVs) which go through a series of maturation steps (Fig. 1.). The biogenesis of granules is beyond the scope of this overview. The V-SNARE synaptobrevin and synaptotagmins are acquired, along with Rab3, synaptophysin, vATPase, and the neurotransmitter transporter (VMAT in the case of chromaffin cells), during trafficking through trans-Golgi and endosomal compartments, where many of these components are recycled. After transport to the plasma membrane, the LDCVs undergo a docking process which anchors them at the plasma membrane. Chromaffin cells provide a model allowing study of the presynaptic processes involved in the release of catecholamines via LDCVs.

Fig. 1:

Synapse function in chromaffin cells of the adrenal gland. The important maturation steps of LDCVs and the proteins involved are shown. LDCVs from the reserve pool in the cytoplasm are transported to the plasma membrane via the cytoskeleton and bind with docking molecules. These docked LDCVs are then primed to fusion competence and finally, following calcium entry, fuse with the plasma membrane and release their cargo of catecholamines.

Docking, Priming and Fusion of LDCVs in Chromaffin Cells

The docking complex is initiated by an interaction between the two SNARE-proteins Syntaxin1 and SNAP-25, which bind to the plasma membrane, and a vesicle-associated protein, Synaptotagmin (de Wit et al., 2009). The availability of Syntaxin1 and thereby the efficiency of the docking process is regulated by Munc18. The docked LDCV is not yet fusion competent and must go through an additional maturation step, the so-called priming step. Munc13s are absolutely necessary for priming at most synapses. They catalyze the assembly of the SNARE complex which is composed of Syntaxin1, SNAP-25 und the vesicle-associated SNARE-protein Synaptobrevin2. Munc13s have recently been proposed to have a priming function in chromaffin cells as well (Man et al., 2015). The assembly of the SNARE complex renders the LDCVs fusion-competent, and after entry of calcium at the presynapse and the binding of calcium to Synaptotagmin, drives the fusion of LDCVs with the plasma membrane resulting in release of catecholamines. Other proteins such as Complexin and CAPS are, in addition, involved in the regulation of docking, priming and fusion of LDCVs in chromaffin cells (Stevens et al., 2011).

Assembling the Immunological Synapse in CTLs

Cytotoxic T-lymphocytes patrol throughout the entire body in search of infectious pathogens. They are transported in the blood stream, penetrate blood vessel walls and migrate through all organs of the body. If CTLs discover infected tissue they can, after activation, kill a large number of target cells, one after the other. This is referred to as “serial killing”. Thus it is obvious that precise recognition of target cells is essential, since CTLs should only kill those cells which present foreign peptides. This requirement leads to one of the great contrasts between chromaffin cells, which produce a large pool of releasable LDCVs, and CTLs, which contain only a few mature CGs which they deliver to the IS on demand.

The assembly of the IS begins with the recognition of foreign antigen on the surface of the target cell by T-cell receptors of the CTL (Fig. 2). The presentation is carried out via the MHC (major histocompatibility complex), which binds peptides resulting from protein degradation by the proteasome, in the endoplasmic reticulum, and presents them after transporting them to the cell surface. The vast majority of presented peptides is not foreign and will not elicit an immune response. The recognition of foreign peptides by T-cell receptors is highly specific and triggers the cascade described below, which ends in the killing of the antigen-presenting cell after release of cytotoxic substances from the CG.

Fig. 2:

Important structures of the Immunological synapse. A After contact with an antigen presenting cell, cytotoxic T-lymphocytes generate an immunological synapse (IS). Shown is a “birds eye view” cartoon of the contact area. The IS can be divided into a peripheral and a central “supramolecular activation cluster”. In the cSMAC there is a secretory zone (SZ) where fusion of cytotoxic granules (CGs) occurs. B CGs are transported along the cytoskeleton to the IS which is built at the contact zone between the CTL and its target. This area is rich in adhesion and signaling molecules (see text). C An electron microscopic image of the contact between CTL and target cell (IS) including two CGs which are being transported to the IS, the centrosomes and Golgi. D Stimulated emission depletion (STED) image of a primary CTL with marked microtubules and centrosome (MTOC).

Binding of the T-cell receptors (TCRs) to antigen leads to a rapid strengthening of the contact due to binding of the adhesion molecules LFA1 (CTL) and ICAM1 (target cell). Thus, as in neurons, the process of adhesion itself plays a major role in generation and maintenance of the contact zone as well as in signaling, although the recognition of target cells in CTLs has little in common with target recognition in neurons. In contrast to neuronal synapse development, there is no requirement for the organization of a postsynaptic response. The induction of cell death follows internalization of the cytotoxic molecule Granzyme, which is facilitated by co-released Granulysins and Perforin.

The “supramolecular activation cluster (SMAC)” forms with TCRs populating the central (cSMAC) area surrounded by adhesion molecules in a ring-shaped peripheral (pSMAC) area. The concentration of TCRs attracts signal molecules including kinases which, by activating phospholipases, lead to production of DAG and IP₃. The IP₃ initiates emptying of the Ca²⁺ stores of the ER which leads to activation of calcium-release activated calcium (CRAC) channels in the plasma membrane. The resulting sustained increase in the intracellular Ca²⁺ concentration initiates an impressive polarization of the CTL and the targeted transport of CGs and other organelles to the IS. Initially the centrosome (referred to as “microtubule organizing center” (MTOC)) moves from its position trailing the nucleus (in migrating CTLs) to a position directly adjacent to the IS. Both actin and tubulin polymerization are required for movement of the MTOC (Fig. 3). Actin polymerization is required for final delivery of CGs to the IS.

Fig. 3:

Synapse assembly, function and disassembly in cytotoxic T-lymphocytes. The assembly of the immunological synapse (IS, left) begins with the fission of recycling endosomes (RE) from the early endosomes (EE). The RE are transported to the IS and fuse (VAMP8-dependent) with the plasma membrane. REs deliver proteins which are required for subsequent docking, priming and fusion of cytotoxic granules (CG). After maturation via late endosomes and lysosomes, CGs (middle) are delivered to the plasma membrane where they anchor (docking). After priming under control of Munc13-4, the SNARE complex drives membrane fusion. Recycling CGs are returned to the cytoplasm via endocytosis (right) where they fuse with the EE and are then trafficked to the LE.

Tubulin polymerization allows projection of the microtubules to the cell interior. Dynein- and Myosin-mediated transport along this cytoskeletal network produces sequential delivery of a variety of vesicles types to the IS. The first to arrive, within the first minute, are Rab11 positive recycling endosomes (RE). Results from our laboratory show that REs are essential for the transport of cargo that is later necessary for the docking, priming and fusion of CGs (Marshall et al., 2015). Among the identified proteins of these REs which fuse with the plasma membrane at the IS in a VAMP8 dependent manner, are Syntaxin11 and Munc13-4. The plasma membrane-bound SNARE and SNARE-associated proteins accumulate in an area of the cSMAC and build the secretory zone (SZ). As is the case in the active zone of neuronal synapses, docking, priming and fusion of CGs occurs exclusively at the SZ. The first CGs reach the IS a few minutes after establishment of contact and the fusion of an average of one to two CGs at the synapse is complete within 15 minutes.

Docking, Priming and Fusion of Cytotoxic Granules in CTLs

Cytotoxic T-lymphocytes (CTLs) are a part of the adaptive immune system, and kill antigen-presenting cells by releasing the cytotoxic proteins Perforin and Granzyme B from cytotoxic granules (CGs). CGs are lysosome-related organelles which are somewhat larger than LDCVs of chromaffin cells and also contain a dense proteinaceous core. Much of what we know about the molecular mechanisms of docking, priming and fusion of CGs is based on lethal immune diseases such as Griscelli Syndrome or familial hemophagocytic lymphohistiocytosis (FHL, (Janka, 2012; de Saint Basile et al., 2010).

Patients with Griscelli Syndrome type 2 have mutations in the GTP-binding protein Rab27a and display albinism and an immune deficit that is fatal if not treated by bone-marrow transplantation. Rab27a interacts with MyosinV in CTLs and is required for correct docking of CGs at the plasma membrane of the IS. In addition, Rab27 interacts with Munc13-4, a member of the Munc13 family. Like its isoforms in the CNS and in neuroendocrine cells, Munc13 is responsible for priming in CGs. Mutation of the Munc13-4 gene leads to the fatal immune disease FHL type 3, since the CGs of these patients are unable to fuse and therefore their CTLs cannot kill target cells (Ménager et al., 2007). Mutations in Munc18-2 also lead to an FHL, in this case type 5, which is the result of the loss of the docking function previously described for Munc18 in neuronal synapses. The fusion of CGs is driven by the SNARE complex, as is the exocytosis of LDCVs in chromaffin cells. In CG fusion a membrane bound SNARE, Syntaxin11, whose mutation leads to FHL type 4, has also been identified via human mutation (Kögl et al., 2013). Further components of the fusion-driving SNARE complex of CGs are Synaptobrevin2 on the CG membrane and probably SNAP-23 in the plasma membrane. In contrast to chromaffin cell synapses, it is still not clear if Synaptotagmin or another calcium binding protein play a role in the actual fusion event. This description of the proteins involved in docking, priming and fusion of CGs indicates their remarkable similarity to the molecular mechanisms at the neuronal synapse (Fig. 3) (Becherer et al., 2012).

Synapse Disassembly

In their function as “serial killers”, CTLs must be able to rapidly disassemble their synapses. The coordinated disassembly of the IS follows within 15-30 minutes after initial contact with the antigen-presenting cell. The lifetime of the immunological synapse is much shorter than that of a neuronal synapse, though in both cases recycling of vesicle components is necessary for function. Though many details are not completely understood, this disassembly can be divided into two processes. T-cell receptors and membrane bound proteins such as Syntaxin11, which are involved in fusion, are internalized via REs and either degraded or delivered to the early endosomes (EE) and recycled.

The membrane components of CGs are also recycled as part of the disassembly process. As we recently demonstrated using a Synaptobrevin2-mRFP knockin mouse, recycling of CG components is independent of the recycling endosomes (Chang et al., 2016). Synaptobrevin2, the vSNARE responsible for CG fusion(Matti et al., 2013) is present in clusters at the plasma membrane immediately after fusion and can be detected with a monoclonal antibody directed at the luminal (intravesicular) mRFP (which is exposed to the extracellular medium at exocytosis, Fig. 4). Within a few minutes, Synaptobrevin2 along with other CG membrane proteins such as the vesicular H⁺-ATPase are endocytosed at the IS. Like the endocytosis of synaptic vesicles in neurons (Soykan et al., 2016) the endocytosis of CG components at the IS depends on Clathrin and Dynamin. In addition, the Synaptobrevin2-specific adaptor protein CALM (Koo et al., 2011) plays an essential role. CGs are acidified after endocytosis and reach the late endosomes via the early endosomes, where they reacquire cytotoxic components such as Granzyme B. Following a series of maturation steps, CGs become available for further fusion at newly built synapses. Approximately 50% of the killing capacity of CGs is achieved by the efficient recycling of CG membrane proteins. The disassembly of synapses including endocytosis is fast enough to allow new IS assembly and killing of antigen presenting cells within a few minutes (Chang et al., 2016).

Fig. 4:

Live visualization of endocytosis of cytotoxic granules. Confocal microscopic images of a cytotoxic T-lymphocyte in contact with an antigen presenting cell (target). The CGs are red due to expression of Synaptobrevin2-mRFP. The chamber medium contains an antibody against mRFP which is bound to a green fluorescent dye. This antibody can only bind synaptobrevin2 after CG exocytosis, when it is exposed on the cell surface. The series begins with contact between CTL and the target cell (0’00’’)`, at time 3’06’’, the IS has formed as evidenced by the polarization of the red CGs to the contact site. In the next image (7’06’’) fusion of CGs has occurred as shown by the green puncta which result from binding of the green-marked mRFP antibody on the red marked Synaptobrevin2-mRFP exposed to the medium (yellow). At the end of the series (13’06’’) several CGs have been endocytosed and transported to the cell interior.

Conclusion

The immunological synapse between cytotoxic T-lymphocytes and antigen-presenting cells has many similarities to neuronal synapses in the CNS. However, in addition to the study of the mechanisms of presynaptic function, in CTLs we have experimental access to the processes of assembly and disassembly, and a system amenable to high-resolution live imaging. Together with the availability of primary human material, these are good arguments for consideration of the immunological synapse in the future as a model system for some aspects of synaptic function.

Abbreviations

CG cytotoxic granule

CRAC calcium release-activated calcium

CTL cytotoxic T-lymphocyte

FHL familial hemophagocytic lymphohistiocytosis

IS immunological synapse

LDCV large dense-core vesicle

MHC major histocompatibility complex

MTOC microtubule organizing center

SMAC supramolecular activation cluster

SNARE soluble N-ethylmaleimide-sensitive factor attachment protein receptor

TCR T-cell receptor

CNS central nervous system