Bassoon and piccolo regulate ubiquitination and link presynaptic molecular dynamics with activity‐regulated gene expression

Abstract

Release of neurotransmitter is executed by complex multiprotein machinery, which is assembled around the presynaptic cytomatrix at the active zone. One well‐established function of this proteinaceous scaffold is the spatial organization of synaptic vesicle cluster, the protein complexes that execute membrane fusion and compensatory endocytosis, and the transmembrane molecules important for alignment of pre‐ and postsynaptic structures. The presynaptic cytomatrix proteins function also in processes other than the formation of a static frame for assembly of the release apparatus and synaptic vesicle cycling. They actively contribute to the regulation of multiple steps in this process and are themselves an important subject of regulation during neuronal plasticity. We are only beginning to understand the mechanisms and signalling pathways controlling these regulations. They are mainly dependent on posttranslational modifications, including phosphorylation and small‐molecules conjugation, such as ubiquitination. Ubiquitination of presynaptic proteins might lead to their degradation by proteasomes, but evidence is growing that this modification also affects their function independently of their degradation. Signalling from presynapse to nucleus, which works on a much slower time scale and more globally, emerged as an important mechanism for persistent usage‐dependent and homeostatic neuronal plasticity. Recently, two new functions for the largest presynaptic scaffolding proteins bassoon and piccolo emerged. They were implied (1) in the regulation of specific protein ubiquitination and proteasome‐mediated proteolysis that potentially contributes to short‐term plasticity at the presynapse and (2) in the coupling of activity‐induced molecular rearrangements at the presynapse with reprogramming of expression of neuronal activity‐regulated genes.

Abbreviations

CAZ

cytomatrix at the active zone

CtBP1

C‐terminal binding protein 1

RIM

Rab3‐interacting molecule

SIAH1

seven in absentia homolog 1

SCF

Skp/Cullin/F‐Box containing

SV

synaptic vesicle

TF

transcription factor

Ub

ubiquitin

UPS

ubiquitin–proteasome system

Neurotransmitter release from presynapses relies on a complex sequence of membrane trafficking processes including translocation of neurotransmitter‐filled synaptic vesicles (SVs) towards the presynaptic membrane, their firm association with the release sites, fusion and release of neurotransmitters, and compensatory endocytosis that is necessary to recycle the SV membrane and protein components. In addition to this vigorous membrane trafficking, the release apparatus, which is an assembly of multiple effector and regulatory proteins, needs to be formed before SV recruitment, and it dissociates after SV fusion (for review see Sudhof, 2012). In spite of the high dynamics of the presynaptic membranes and proteins during neurotransmission, synaptic connections of neurons in vitro and in vivo do not lose their integrity and remain stable over days, years or possibly even lifetimes. The structure that was suggested to confer stability to the presynaptic molecular assemblies is the so‐called cytomatrix at the active zones (CAZ). The core components of the cytomatrix are liprin‐α, Munc13, Rab3‐interacting molecules (RIMs), RIMs‐binding proteins (RBPs), ELKS proteins (rich in the amino acids E, L, K and S; also called CAST) and bassoon/piccolo. Common features of these proteins are their relatively large size and multi‐domain structure. This allows them to have multiple interacting partners, including binding with each other, and thereby to form a large multi‐molecular meshwork, which functions in scaffolding and in the restriction of membrane trafficking events at the presynaptic active zones of neurotransmitter release (reviewed in Gundelfinger & Fejtova, 2012; Gundelfinger et al. 2015; Michel et al. 2015). Apart from their scaffolding function, the CAZ proteins are also important regulators of neurotransmission via the regulation of SV tethering and priming, assembly of release sites, including the recruitment of voltage‐gated calcium channels, and spatial and functional coupling of endo‐ and exocytotic machinery (for reviews see Gundelfinger & Fejtova, 2012; Ackermann et al. 2015).

Despite their importance in the spatial and temporal stabilization of presynaptic release machinery, live imaging studies using fluorescently tagged CAZ molecules demonstrated that these molecules are highly dynamic in presynaptic terminals (Kalla et al. 2006; Tsuriel et al. 2009; Schroder et al. 2013; Spangler et al. 2013). It is still a matter of debate as to whether this represents the mere stochastic dynamics characteristic of any living system or whether this reflects/drives changes in presynaptic efficacy (reviewed in Gundelfinger & Fejtova, 2012; Ziv & Fisher‐Lavie, 2014; Michel et al. 2015). A growing body of evidence is in support of the latter. Posttranslational modifications, new protein synthesis and degradation and reconfiguration of gene expression seem to contribute to the molecular remodelling at the presynapse, whilst functioning on different temporal and spatial scales (Abstract figure). Several examples demonstrate that transient posttranslational modifications of the presynaptic scaffolds can rapidly modulate the composition of the presynaptic release machinery. They can act locally and are probably the initial event in alteration of presynaptic efficacy, having therefore potential to contribute to the short‐term plasticity at individual presynapses (Abstract figure). It has been shown that phosphorylation modulates the anchoring of bassoon to the presynaptic CAZ (Schroder et al. 2013), or clustering of Munc18‐1 at release sites (Cijsouw et al. 2014). Phosphorylation of synapsin I at multiple sites regulates the dynamics and functional properties of SVs and affects the presynaptic function and plasticity (Chi et al. 2001; Menegon et al. 2006; Verstegen et al. 2014). Recently, an attachment of a small Ub‐related modifier (SUMO) to RIM was suggested to modulate its role in the recruitment of the presynaptic N‐type voltage‐gated channel (Girach et al. 2013). Dynamic ubiquitination of CAZ proteins, as will be discussed in detail later, has emerged as an important, rapid mechanism in the modulation of presynaptic release efficacy. Ubiquitination is also known to target proteins for degradation via the ubiquitin–proteasome system (UPS), which together with local protein synthesis, represents yet another mechanism that can contribute to the remodelling of the molecular composition at presynapse. These mechanisms, however, are slower and with a lower local specificity than the rapid effects of the aforementioned posttranslational modifications (Abstract figure) and seem to be important during neuronal development, neuronal regeneration and for some postsynaptic forms of synaptic plasticity (Jung et al. 2012; Hegde et al. 2014). At the presynapse, UPS‐dependent degradation has been implicated in the remodelling of the presynaptic cytomatrix during homeostatic adaptations of the presynaptic efficacy to changes in global neuronal network activity levels as reviewed elsewhere (Lazarevic et al. 2013).

Two recent findings add an important new facet to the complex picture of mechanisms of regulation of the molecular composition of the CAZ. The largest scaffolds of the CAZ, bassoon and piccolo, have been shown to contribute to the shaping of the presynaptic composition by two distinct mechanisms: the local regulation of the UPS at presynapse and the linking of presynaptic activity to neuronal gene expression reprogramming, which affects the proteome at the level of whole neuron. The latter acts on a much longer time scale (Abstract figure) and probably contributes to the persistent changes in the presynaptic function during usage‐dependent neuronal plasticity.

Role of the presynaptic UPS in the regulation of neurotransmitter release

The manipulation of ubiquitination and/or proteasome activity has a rapid effect on neurotransmitter release from the presynapse in both vertebrates and invertebrates. More than a decade ago it was shown that inhibition of the proteasomal catalytic activity by lactacystin or epoxomycin induced a rapid (within 45 min) increase in neurotransmission at Drosophila neuromuscular junction (NMJ) (Speese et al. 2003). An identical treatment increased the synaptic abundance of presynaptic Dunc13, while other targets of UPS‐mediated degradation such as syntaxin, synaptotagmin or Leonardo/14‐3‐3ζ were unaffected. This led to the hypothesis that inhibition of the UPS‐dependent degradation of Dunc13 mediates the observed effects. In line with this prediction, a fly mutant of the β6 subunit of the core proteasome, which has a decreased proteasomal catalytic activity, also showed an accumulation of Dunc13 at the NMJ (Speese et al. 2003). In cultured hippocampal neurons, treatment with MG132 or clasto‐lactacystin β‐lactone for just a few minutes induced an up to fourfold increase in spontaneous neurotransmission (Rinetti & Schweizer, 2010). However, in contrast to the situation in Drosophila, the synaptic abundance of Munc13 and its regulator RIM were unaltered by this treatment. This fact, together with the reported half‐lives of Munc13 of >1 day in vitro (Cohen et al. 2013) and about 10 days in vivo (Price et al. 2010), and the finding that inhibition of the proteasomal activity has no effect on its abundance or dynamics (Kalla et al. 2006), suggest that dynamic protein ubiquitination and consequent signalling (Hallengren et al. 2013), and not proteolysis of target proteins per se, are absolutely crucial. Indeed, pharmacological inhibition of E1 ligase activity, which results in reduced protein ubiquitination, increased neurotransmission in the same way as proteasome inhibition (Rinetti & Schweizer, 2010).

Studies on the neuronal phenotypes of spontaneously occurring ax(J) mice that carry a mutation in the Usp14 gene and show a progressive motor impairment with ataxia pointed in the same direction (Wilson et al. 2002). Usp14 is a proteasome‐associated ubiquitin (Ub)‐specific protease that trims Ub‐chains on proteins before their further processing by the proteasome, which allows recycling of mono‐Ub necessary for effective dynamic protein ubiquitination. However, Usp14 can also remove Ub moieties from certain proteins to rescue them from UPS‐mediated degradation (for review see Ristic et al. 2014). The ax(J) mice show a defect in synaptic transmission, which is due to a reduced facilitation of vesicular release probability during repetitive stimulation (Walters et al. 2014). Normal proteasomal activity, but reduced free Ub levels, were measured in brain lysates or synaptosomes of ax(J) mice (Anderson et al. 2005; Chen et al. 2009, 2011). Though the molecular target of Usp14 action is not completely understood, modulation of the CAZ proteins could explain the observed changes. Walters and coauthors did not find any changes in the total abundance of several presynaptic proteins including Munc13 and RIM in ax(J) mice (Walters et al. 2014). However, it remains to be determined to what extent Usp14 regulates their dynamic ubiquitination and consequent monoUb‐conjugation‐induced signalling. Munc13 and RIM are ubiquitinated by the F‐box proteins Fbxo45 and Fbxl20 (also known as scrapper) (Yao et al. 2007; Tada et al. 2010), which are the substrate‐specificity determining parts of modular E3 ligase complexes similar to SCF (Skp/Cullin/F‐Box containing) (Jin et al. 2004; Saiga et al. 2009). In neurons, the functions of these E3 ligase complexes are not restricted to regulation of neurotransmission. Fbxo45 (and its worm and Drosophila homologues FSN‐1 and DFsn, respectively) act in a complex with E3 ligase PHR‐1 (known as RPM‐1 in worms and Hiw in Drosophila), which play a role in the control of axonal growth and pathfinding, in synapse formation and in injury‐induced retrograde signalling, as described later (for review see Tian & Wu, 2013). Fbxl20/scrapper was implied in the control of receptor endocytosis and autophagy in non‐neuronal cells (Kuchay et al. 2013). A depletion of either Fbxo45 or Fbxl20 in mature hippocampal neurons led to an increase in spontaneous transmission, while their overexpression had the opposite effect (Yao et al. 2007; Tada et al. 2010). Moreover, inhibition of the proteasome by MG132 or epoxomycin had no effect in Fbxl20/scrapper mice (Yao et al. 2007). All of these findings support the notion that RIM and Munc13 are the molecular substrates for neurotransmission regulation by the proteasome.

Prolonged inhibition of the proteasome by pharmacology or by genetic interference induces an increase in the recycling synaptic vesicle pool. This effect requires inhibition of the proteasomal activity for at least 1 h and seems to be distinct from the rapid effect on SV release probability, which is evident within a few minutes (Willeumier et al. 2006). This regulation of the recycling SV pool is activity dependent: proteasome inhibition has no effect in cells upon activity withdrawal, but is evident already after 15 min in cells with increased network activity. This suggests that prolonged inhibition of the proteasome affects the correct membrane sorting or trafficking. Interestingly, this effect seems to be dependent on cAMP–protein kinase A (PKA) signalling and can be blocked by pharmacological inhibition of this pathway (Willeumier et al. 2006). SV pools are also affected in ax(J) mice, which have a lower number of docked and total vesicles per synapse (Walters et al. 2014). The downstream molecular targets of the UPS‐mediated regulation of the SV pools are unclear. The SV‐associated scaffolding protein synapsin, which is an important PKA‐dependent regulator of the SV pools (Menegon et al. 2006), might be a good candidate.

Very little is known about the mechanisms of activity‐dependent control of ubiquitination and proteasomal activity at the presynapse. A recent study demonstrated that the presynaptic scaffolding proteins are not only targets of ubiquitination, but also important regulators of ubiquitination and degradation of synaptic proteins (Waites et al. 2013). Waites and colleagues showed that the large presynaptic scaffolding proteins bassoon and piccolo can bind to and negatively regulate the activity of the RING domain‐containing E3 ubiquitin ligase SIAH1, a member of the seven in absentia homolog family. SIAH1 acts in concert with an adenomatous polyposis coli‐based SCF E3 ligase complex to induce proteolysis of the important signalling molecule β‐catenin (Dimitrova et al. 2010) and also ubiquitinates several neuronal proteins including the SV protein synaptophysin (Wheeler et al. 2002). Neurons deficient for both bassoon and piccolo showed a progressive degenerative phenotype: SVs were enlarged and abnormal endo‐lysosomal structures and multivesicular bodies accumulated in the synapses and cell bodies. These phenotypes could be attenuated by proteasomal or lysosomal inhibition or by genetic interference with SIAH1 expression suggesting that dysregulation of SIAH1‐dependent proteostasis accounts for the described phenotype (Waites et al. 2013). SIAH1‐induced ubiquitination not only targets proteins for degradation, but also regulates multiple neuronal signalling pathways, including Wnt/β‐catenin, TNF‐α and p53‐based apoptotic signalling, which lead to the reprogramming of cellular expressional patterns. It will be interesting to investigate in future studies the role of the functional interaction between SIAH and bassoon/piccolo in SIAH‐related signalling during synaptic plasticity.

Presynapse‐to‐nucleus signalling mediated by the transcriptional co‐repressor CtBP1

CtBP1 belongs to the family of C‐terminal binding proteins, which are well‐characterized ubiquitous negative transcriptional regulators. They act by recruiting histone deacetylases, methyltransferases and demethylases to transcriptional complexes and by inhibiting histone acetyltransferases, both of which lead to chromatin compaction and gene silencing (Chinnadurai, 2007). CtBP1 was also linked to regulations of membrane fission at trans‐ and cis‐Golgi and at the plasma membrane (Valente et al. 2013). However, this function seems to be not required for neuronal‐specific functions (Vaithianathan et al. 2013). CtBPs are nuclear and cytoplasmic proteins in most cell types, but in neurons they are also highly enriched in the presynapse (tom Dieck et al. 2005; Hubler et al. 2012; Ivanova et al. 2015). Ribeye, a member of the CtBP family that is expressed exclusively in the retinal photoreceptor and bipolar cells, in the hair cells of the cochlea and in the pinealocytes, is the only exception: it shows no nuclear localization and is restricted to the presynaptic specialization in these cells, the so‐called presynaptic ribbon (Schmitz et al. 2000; for review see Zanazzi & Matthews, 2009). CtBP1 is localized to presynapses via its association with the CAZ scaffolding proteins bassoon and piccolo. In the absence of both bassoon and piccolo the presynaptic enrichment of CtBP1 is lost (tom Dieck et al. 2005; Ivanova et al. 2015). Recently, Ivanova and colleagues showed that neuronal activity regulates the distribution of CtBP1 between the presynapse and the nucleus and thereby controls CtBP1‐dependent transcriptional repression of several important neuronal activity‐regulated genes (Fig. 1). In highly active neurons, CtBP1 protein abundance is increased at presynapses and decreased in nuclei: transcriptional repression is released, while upon silencing of the network activity, CtBP1 decreases at presynapses and accumulates in nuclei where it represses gene transcription. Although some mechanisms underlying shuttling of CtBP1 between the cytoplasm and nucleus might be conserved between neurons and non‐neuronal cells, the contribution of the synaptic pool of CtBP1 seems to be crucial for neuronal activity‐induced reprogramming of gene expression. Indeed, in neurons, where synaptic CtBP1 was ablated, activity failed to regulate its nuclear abundance and the consequent induction of activity‐regulated genes (Ivanova et al. 2015). The shuttling of CtBP1 from presynapses to nucleus is a novel example of an activity‐dependent communication between the distal axons and the somatic compartment. It is reminiscent of, and might share some common mechanisms with, two distinct pathways that were previously shown to connect these compartments: signalling mediated by a retrograde translocation of activated transcription factors (TFs) or transcriptionally active multimolecular complexes in axonal outgrowth during development and after injury (Panayotis et al. 2015) and retrograde transport of signalling endosomes (Schmieg et al. 2014). The injury‐related signalling complexes and the signalling endosomes are linked to the dynein molecular motors that mediate their retrograde transport along microtubular tracks. Shuttling of CtBP1 between distal axons and nucleus occurs in a timeframe of hours and is therefore likely to be dependent upon active axonal transport. CtBP1 is anterogradely transported and associated with transport vesicles named piccolo/bassoon transport vesicles, which mediate the anterograde transport of active zone components to distal axons. However, the mode of retrograde transport of CtBP1 is unclear. For axon injury‐induced retrograde signalling of TF, importin‐α/β can function as a cargo adaptor for the retrograde transport of transcriptionally active signalling complexes (Rishal & Fainzilber, 2014). Interestingly, a macromolecular complex containing CtBP1 and importin‐α was identified in a non‐neuronal context (Yousef et al. 2012), which allows speculation that this transport mechanism can be shared between CtBP1 and injury signalling pathways (Fig. 1 B). The signalling endosomes are formed by endocytosis of transmembrane neurotrophin receptors after ligand binding and mediate a switch from a local signalling around the ligand‐bound receptor at the membrane to a long‐range signalling by this transport organelle (Cosker & Segal, 2014; Matusica & Coulson, 2014). In axonal outgrowth and in axonal injury signalling, the retrograde transport of an importin‐based cargo is induced by assembly of transcriptionally active multimolecular complexes, which require local protein synthesis of transcription factors (e.g. STAT3 or SMAD) and cargo adapters such as importin‐β (Ji & Jaffrey, 2014). Initiation of retrograde signalling via the JUN kinase pathway involves balanced ubiquitination/deubiquitination of specific targets (e.g Map3k12, also named dual leucine zipper kinase DLK) that subsequently regulate retrograde transport of the Jun amino‐terminal kinase 3 (JNK3)‐containing cargo (reviewed in Tian & Wu, 2013). The CtBP1 signalling does not require local protein synthesis or protein degradation. Instead, neuronal activity rapidly regulates the dynamic association of CtBP1 with presynapses (Ivanova et al. 2015). Increased neuronal activity leads to a drop in the NAD/NADH ratio and because CtBP1 bound to NADH shows a much tighter association with bassoon this leads to its increased synaptic association. This postulates the CtBP1 signalling as a new mechanism linking activity‐induced molecular rearrangements of the CAZ with reconfiguration of the expression of activity‐regulated genes. All data about the activity‐dependent regulation of the CtBP1 signalling pathway were obtained from experiments in which the global network activity was manipulated. It will be interesting to test whether this type of presynapse‐to‐nucleus signalling can also transduce signals related to potentiation of a specific synaptic contact in intact circuits and thereby contribute to input‐specific neuronal plasticity.

Figure 1. CtBP1 signalling pathway .

A, at the presynapse CtBP1 is anchored via its direct interaction with the CAZ proteins bassoon and piccolo. This interaction is sensitive to the cellular NAD/NADH ratio, which is altered by neuronal activity. In conditions of increased neuronal activity the NADH concentration increases, which leads to a firmer association of CtBP1 with the CAZ and its concomitant nuclear depletion leads to a release of the transcriptional repression on activity‐regulated genes. In conditions of low activity, CtBP1 dissociates from the CAZ and at the same time its levels are increased in cell soma/nucleus. This process probably involves active transport (B) analogous to the importin‐mediated axonal retrograde trafficking of transcription factors (TF) that mediates signalling during axonal outgrowth in development and after injury. Once CtBP1 translocates to the nucleus (C) it acts there to inhibit transcription of activity‐regulated genes.

Implication of CtBP1 signalling in brain disease

The distribution of CtBP1 between synapses and nucleus is influenced by the cellular metabolic status. In neurons in which the metabolism of glucose (glycolysis) is inhibited, the activity‐regulated shuttling of CtBP1 between synapses and nucleus, and the ensuing control of activity‐induced gene expression by neuronal activity, are disturbed (Garriga‐Canut et al. 2006; Ivanova et al. 2015). The ketogenic diet, which is characterized by a low carbohydrate content and hence reduced glycolysis, proved to be effective in the therapy of treatment‐resistant forms of epilepsy, where pathologically increased neuronal activity leads to unconstrained activation of gene expression (Garriga‐Canut et al. 2006; Lutas & Yellen, 2013). The release of CtBP1‐mediated transcriptional repression in conditions of high neuronal activity, and the fact that inhibition of glycolysis diminished the activity‐dependent shuttling of CtBP1, suggests that the CtBP1‐mediated control of activity‐regulated genes might play a crucial role in the pathology of seizures (Garriga‐Canut et al. 2006; Ivanova et al. 2015). This also highlights the importance of the investigation of the molecular mechanisms that link synaptic activity with regulation of gene expression for our understanding of the function of the healthy brain as well as for the development of new approaches for the therapy of brain diseases.

Additional information

Competing interests

None declared.

Funding

A.F. was supported by DFG FE 1335‐1, CBBS and Bundesministerium für Forschung und Technologie (BMBF). D.I and A.F. were supported by Leibniz Association (Pakt für Forschung und Innovation), D.I. and A.D. by DFG/GRK 1176.

Biographies

Daniela Ivanova recently obtained her PhD with Professor Eckart D. Gundelfinger and Dr Anna Fejtova in the Presynaptic Plasticity group at the Leibniz Institute for Neurobiology (LIN) in Magdeburg for investigations of the role of CtBP1 in activity‐dependent gene regulation and synaptic vesicle trafficking in neurons.

Anika Dirks also works in the lab of Eckart D. Gundelfinger and Anna Fejtova at LIN. Her current PhD project focuses on the regulation of the signalling messenger CtBP1 in neurons.

Anna Fejtova obtained her PhD at the Institute of Biochemistry, University of Zurich and Neuroscience Centre Zurich (Switzerland). She joined the laboratory of Eckart D. Gundelfinger at LIN as a postdoctoral fellow of the Swiss National Foundation to investigate the synaptic functions of presynaptic scaffolds bassoon and piccolo. Currently, she leads the Junior Research Group Presynaptic Plasticity at LIN that focuses on investigations of molecular mechanisms that mediate the presynaptic functional remodelling during experience‐induced, homeostatic and pathological neuronal plasticity.