Ubiquitin ligase signaling networks shape presynaptic development, function and disease

Abstract

Ubiquitin ligases are important regulators of nervous system development, function and disease. To date, numerous ubiquitin ligases have been discovered that regulate presynaptic biology. Here, we discuss recent findings on presynaptic ubiquitin ligases that include members from the three major ubiquitin ligase classes: RING, RBR and HECT. Several themes emerge based on findings across a range of model systems. A cadre of ubiquitin ligases is required presynaptically to orchestrate development and transmission at synapses. Multiple ubiquitin ligases deploy both enzymatic and non-enzymatic mechanisms, and act as hubs for signaling networks at the synapse. Both excitatory and inhibitory presynaptic terminals are influenced by ligase activity. Finally, there are several neurodevelopmental disorders and neurodegenerative diseases associated with presynaptic ubiquitin ligases. These findings highlight the growing prominence and biomedical relevance of the presynaptic ubiquitin ligase network.

Graphical Abstract

Depicted is the ubiquitin ligase network that positively (arrows) and negatively (bars) affects different processes that influence presynaptic development and function.

Introduction

Ubiquitin ligases play prominent roles in nervous system development, function and disease (Yamada et al., 2013; Kawabe & Stegmuller, 2021; Pinto et al., 2021). There are three major ubiquitin ligase classes including Really Interesting New Gene (RING), Homologous to E6-AP C-terminus (HECT), and RING between RING (RBR) ligases, all of which have extensive roles in the nervous system. These classifications are primarily based on structural and biochemical studies (Zheng & Shabek, 2017; Harper & Schulman, 2021). RING, RBR and HECT class ubiquitin ligases are conserved from simple invertebrates like C. elegans and Drosophila through humans. RING ubiquitin ligases function as multi-subunit complexes, while HECT ubiquitin ligases are single subunit enzymes. RBR ubiquitin ligases display a hybrid RING-HECT ubiquitination mechanism. In general, polyubiquitination of substrate proteins leads to inhibition and degradation via the proteasome, while mono-ubiquitination results in altered protein trafficking and function. There is now substantial evidence that both RING and HECT ubiquitin ligases interact with binding proteins to engage in signaling activities that are independent of enzymatic ligase activity. This highlights the importance of non-enzymatic functions and has identified ubiquitin ligases that act as signaling hubs with both non-enzymatic and enzymatic functions.

There has been progress identifying numerous ubiquitin ligases that affect presynaptic development and function. Some of the ligases we discuss here include: human MYCBP2 (also called Phr1 in mice, Highwire in Drosophila and RPM-1 in C. elegans), SIAH1, Parkin (also called PARK2), UBE3A, HERC1, and HUWE1 (also called EEL-1 in C. elegans). Interest in presynaptic ubiquitin ligases is heightened by clinical studies indicating many are involved in neurological disorders and diseases. Importantly, there is evidence some ligases anchor presynaptic signaling networks, which are only beginning to be deciphered. Here, we discuss recent developments in our understanding of presynaptic ubiquitin ligases, principally focusing on discoveries from the past 5 years. Earlier studies on other important presynaptic ubiquitin ligases such as Staring, Scrapper and APC are covered in prior reviews (Folci et al., 2020; Kawabe & Stegmuller, 2021).

Siah1 and Parkin regulate presynaptic autophagy, mitochondria and synaptic vesicle dynamics

Proteostasis of presynaptic machinery is critical for the development and function of post-mitotic neurons. At the presynaptic terminal, proteostasis plays important roles in recycling and replacing components of the active zone, synaptic vesicles and organelles. Both the ubiquitin proteasome and autophagy are utilized to overturn presynaptic machinery. Two ubiquitin ligases, Parkin and Seven in absentia homolog 1 (Siah1), have been identified as key regulators of both synaptic vesicle dynamics and presynaptic autophagy.

Siah1 is a RING ubiquitin ligase in the seven in absentia homolog family that has links to both neurodevelopmental and neurodegenerative disorders. Connections between neurodegenerative disease and Siah1 have been found for Parkinson’s disease, as SIAH1 ubiquitinates alpha-synuclein and is detected in Lewy bodies (Yoo et al., 2022). Regarding clinical neurodevelopmental observations, de novo variants for SIAH1 have been identified in multiple patients with developmental delay, hypotonia, and dysmorphic features (Buratti et al., 2021). Outcomes with cell-based Wnt/Axin signaling assays indicate that patient-associated variants alter SIAH1 function and are likely pathogenic. Studies using hippocampal neurons have shown Siah1 regulates the synaptic vesicle pool and synaptic vesicle recycling (Waites et al., 2013). Siah1 overexpression was sufficient to reduce the synaptic vesicle pool and increase presynaptic lysosomes and multi-vesicular bodies. Pharmacological results indicated that Siah1 regulates synaptic vesicle dynamics via both the proteasome and endo-lysosomal degradation. These findings are consistent with the earlier observation that Siah1 ubiquitinates and degrades the synaptic vesicle protein Synaptophysin (Wheeler et al., 2002). Interestingly, Siah1 is regulated at presynaptic terminals by the active zone scaffolds Bassoon (Bsn) and Piccolo (Pclo) (Waites et al., 2013). Pclo and Bsn were found to bind Siah1, and synaptic degradation caused by Pclo/Bsn double knock down was suppressed by reducing Siah. These results indicate that Siah1 is inhibited by binding Pclo/Bsn. At present, substrate(s) that Siah1 ubiquitinates to regulate the synaptic vesicle pool remains unclear.

Further studies showed that Bsn inhibits autophagy at the presynaptic terminal (Okerlund et al., 2017). Bsn/Pclo double knockdown resulted in an increase in autophagic vesicles at presynaptic boutons of hippocampal neurons, which occurred alongside loss of synaptic vesicles. This autophagic effect is principally mediated by Bassoon, as single knockdowns showed increased presynaptic and axonal accumulation of autophagosomes. While polyubiquitination was found to regulate presynaptic autophagy, this did not occur though Siah1. Instead, an E3 ligase-like protein, Atg5, was required for autophagy, which was bound and inhibited by Bassoon. Knocking down Atg5 rescued reduced synaptic vesicle levels and increased autophagosome levels in Bsn/Pclo double knockdown neurons.

Parkin is an RBR ubiquitin ligase that is heavily linked to Parkinson’s Disease (Dawson & Dawson, 2010). While the role of Parkin in ubiquitinating proteins at postsynaptic terminals is more extensively studied, it also localizes to and affects presynaptic processes (Sassone et al., 2017). For example, we have known for over two decades that Parkin ubiquitinates presynaptic proteins, such as Synphilin-1 and Endophillin-A (Chung et al., 2001; Ribeiro et al., 2002; Trempe et al., 2009). The importance of presynaptic Parkin function is highlighted by human brain imaging studies that indicate patients with mutations in Parkin/PARK2 have presynaptic alterations in dopaminergic neurons (Hilker et al., 2001; Ribeiro et al., 2009). Recent mechanistic insight into the presynaptic role of Parkin emerged when it was shown to function with Bsn to regulate presynaptic vesicle turnover via autophagy (Hoffmann-Conaway et al., 2020; Montenegro-Venegas et al., 2020). Proteomics with neurons derived from knockout mice showed that loss of Bsn results in increased ubiquitination of several presynaptic proteins (SV2, Vamp, Syntaxin and Synaptotagmin) (Hoffmann-Conaway et al., 2020). Supporting these molecular observations, Bsn knockout neurons showed increased synaptic vesicle turnover rates, which were accompanied by increased presynaptic autophagosome accumulation assessed via electron microscopy (EM). In contrast to Bsn inhibiting presynaptic autophagy, Parkin promotes autophagy as Parkin knockdown reduced presynaptic levels of the autophagosome marker LC3. Importantly, Parkin knockdown suppressed both synaptic vesicle defects and presynaptic autophagosome accumulation caused by loss of Bsn, which indicates Bsn inhibits Parkin. Collectively, these studies have shown that Parkin, SIAH1, and Atg5 are a group of ligases that regulate presynaptic proteostasis and autophagy to influence synaptic vesicle turnover and recycling (Fig 1).

Fig. 1. Siah1 and Parkin drive presynaptic autophagy, SV dynamics, and endosomal sorting.

Studies with hippocampal neurons indicate that Siah1 and Parkin regulate synaptic vesicle recycling and autophagy at the presynaptic terminal. The active zone scaffolding proteins Bassoon and Piccolo inhibit Siah1 and Parkin. Parkin is a broad acting presynaptic ligase that regulates autophagy, synaptic vesicle dynamics, endosomal sorting, and mitochondria via poly- and mono-ubiquitination of multiple substrates.

Another function of Parkin is regulation of mitochondrial transport and mitophagy, a specific form of autophagy that degrades mitochondria (Han et al., 2023). Parkin, activated by PINK1, promotes degradation of damaged mitochondria through ubiquitination of mitochondrial proteins and recruitment of autophagic machinery. The potential significance of Parkin-mediated mitophagy on presynaptic function was recently supported by observations suggesting Parkin is required for axonal trafficking of mitochondria and mitophagy in hippocampal neurons derived from a transgenic Tau P301L mouse model (Jeong et al., 2022). Pharmacologically activating Parkin increased LC3 levels and recruitment of LC3 to mitochondria, two hallmarks of mitophagy. Furthermore, Parkin knockdown rescued reduced axonal trafficking of mitochondria in Tau P301L neurons. These observations suggest that altered Parkin activity could mediate reduced presynaptic mitochondrial accumulation observed in AD patient brains and the Tau P301L model (Jeong et al., 2022).

Recently, a study using Parkinson’s Disease patient-derived, dopaminergic neurons provided further evidence that Parkin affects synaptic vesicle recycling (Song et al., 2023). EM imaging showed Parkin mutant neurons have fewer, larger synaptic vesicles and increased clathrin at presynaptic membranes. These results indicate that clathrin-mediated endocytosis increases when Parkin-mediated synaptic vesicle recycling is impaired. Parkin effects on vesicle recycling occurred partially through ubiquitination of Synaptojanin-1, which influences membrane trafficking and lipid homeostasis (Choudhry et al., 2021; Song et al., 2023). Interestingly, Parkin is activated by CaMKII kinase. Activated Parkin then ubiquitinates Synaptojanin-1 to promote downstream binding of Synaptojanin with Endophilin A1 (Song et al., 2023) (Fig 1). Thus, a CaMKII/Parkin/Endophilin signaling axis regulates synaptic vesicle recycling in primary human neurons. Examining this signaling axis using in vivo models, while challenging, will be an important next step for the field.

Finally, we highlight that Parkin effects on endosomal sorting could have potential implications for presynaptic function. Parkin has been shown to affect several molecules and complexes involved in the retromer, which plays a key role in endosomal sorting. First, Parkin ubiquitinates the retromer component VPS35 (Williams et al., 2018). Second, the brains of Parkin mutant mice display reduced levels of the WASH complex, which is involved in retromer trafficking (Gomez & Billadeau, 2009; Williams et al., 2018). Consistent with this, sorting of ATG-9 (a WASH-dependent retromer cargo) is disrupted by Parkin knockdown in cultured neurons (Williams et al., 2018). Lastly, Parkin ubiquitinates Rab7a, which recruits the retromer to endosomes (Song et al., 2016). Thus, Parkin potentially influences the retromer and endosomal sorting via multiple molecular interfaces. We draw particular attention to Parkin effects on VPS35, as studies in Drosophila dopaminergic neurons indicate VPS35 regulates synaptic vesicle release and recycling (Inoshita et al., 2017).

Taken as a whole, this body of work provides substantial evidence that Parkin and Siah1 could function to regulate a network of presynaptic players and coordinately influence autophagy and synaptic vesicle recycling. This model is consistent with both ligases sharing a common Bsn/Pclo upstream regulatory mechanism. Dissecting how this potential ligase network influences presynaptic stability and function remains an important future direction given the links between both ligases and neurological disease.

Ube3A and Ube3B shape presynaptic development and function in mammals and Drosophila

Ube3A is a HECT ubiquitin ligase that regulates both glutamatergic and GABAergic neuronal function with many early studies focused primarily on Ube3A postsynaptic functions (Margolis et al., 2010; Smith et al., 2011; Wallace et al., 2012; Khatri & Man, 2019). UBE3A deficiency causes the neurodevelopmental disorder Angelman Syndrome, and UBE3A copy number variants are linked to autism (Kishino et al., 1997; Matsuura et al., 1997; Glessner et al., 2009; Williams et al., 2010; Mabb et al., 2011). The importance of Ube3A at the presynaptic terminal was demonstrated in studies on mice where maternal Ube3A was globally eliminated (Ube3A ^m-/p+) (Wallace et al., 2012). Electrophysiological results with this Angelman syndrome model indicated that presynaptic transmission and synaptic vesicles were reduced in both excitatory and inhibitory neurons. The relevance of Ube3A to presynaptic function is further supported by immunohistochemistry with electron microscopy that showed Ube3A is enriched at glutamatergic and GABAergic presynaptic terminals of hippocampal and cortical neurons in both mice and humans (Burette et al., 2017; Burette et al., 2018). Subsequent studies on the Calyx of Held auditory synapse showed that Ube3A mutant mice where Ube3a is globally impaired results in enhanced presynaptic transmission (Wang et al., 2017). Taken together, these findings suggest that a balance of Ube3A activity could be required for presynaptic transmission in inhibitory GABA and excitatory glutamatergic neurons in the brain. At present, the molecular mechanism for how Ube3A influences presynaptic transmission remains unknown. While Ube3A enzymatic HECT ligase activity could be involved in presynaptic transmission, this has not been tested with targeted genetic approaches that specifically impair ubiquitin ligase activity and Ube3A substrates that mediate effects on presynaptic transmission remain to be identified (Fig 2A).

Fig. 2. Ube3A and Ube3B regulate presynaptic pruning and function.

A) Summary of Ube3A function in mammalian neurons. Ube3A promotes excitatory synaptic transmission, and inhibits clathrin-coated vesicle accumulation in inhibitory neurons. B) At the fly NMJ, Ube3A promotes synaptic pruning via polyubiquitination and inhibition of Tkv thereby restricting a Tkv/Mad signaling axis. C) In rodents, Ube3B (a paralog of Ube3A) promotes Synapsin I levels in excitatory neurons and restricts VGAT levels in inhibitory neurons. This suggests Ube3B could influence excitatory and inhibitory presynaptic function. Ube3B polyubiquitinates the BCKDK kinase and Ppp3cc phosphatase subunit, but it is unclear if Ube3B functions through these molecular mechanisms to influence presynaptic transmission.

In Drosophila, Ube3A was shown to function presynaptically in motor neurons to regulate synapse growth at the neuromuscular junction (NMJ) (Li et al., 2016). In vivo genetic, biochemical and cell biological results indicate this occurs via binding, ubiquitination and proteasome-mediated degradation of the Thickvein (Tkv) BMP receptor, which then affects downstream Mad phosphorylation and function. More recently, this Ube3A/Tkv/Mad signaling axis was found to affect presynaptic pruning in fly mechanosensory neurons (Furusawa et al., 2023). Ube3A regulates pruning by inhibiting the Tkv substrate and regulating developmental changes in Tkv levels that occur during larval to pupal transition. Ube3A specifically affects presynaptic pruning, as dendritic pruning was not altered in Ube3A mutants. Conversely, Ube3A overexpression caused excessive pruning, which reduced synaptic transmission. Kinesin was found to regulate trafficking of Ube3A to presynaptic terminals. Interestingly, engineering Angelman-associated missense variants into flies resulted in decreased synaptic pruning, reduced colocalization of axonal Kinesin with Ube3A, and disrupted Ube3A transport to presynaptic terminals. Overexpression of the Ube3A substrate Tkv phenocopied Ube3A loss of function effects on presynaptic pruning (Fig 2B). Further roles for ubiquitin ligases in presynaptic pruning emerged from Drosophila where a RING ubiquitin ligase, Highwire/MYCBP2, was found to regulate axon pruning in the giant fiber (Borgen et al., 2017a). Like Ube3A, human genetic variants in MYCBP2 were recently shown to cause a neurodevelopmental spectrum disorder (AlAbdi et al., 2023). Thus, multiple ubiquitin ligases linked to neurodevelopmental disorders have been shown to affect presynaptic/axonal pruning. Furthermore, results from invertebrate models such as Drosophila and C. elegans have proved instrumental in providing key insight into the pathogenicity and molecular genetic basis of these neurodevelopmental disorders.

Like its paralog Ube3A, Ube3B is also a HECT ubiquitin ligase with roles in dendritic and postsynaptic development (Cheon et al., 2019; Ambrozkiewicz et al., 2021). There are also observations suggesting that Ube3B plays a role in presynaptic development and transmission. Understanding both pre and postsynaptic functions of Ube3B is important as genetic variants in UBE3B cause a neurodevelopmental disorder, Kaufman oculocerebrofacial syndrome (KOS) (Basel-Vanagaite et al., 2012; Flex et al., 2013). Intellectual disability and developmental delay are prominent neurobehavioral characteristics of KOS. Studies on primary cortical neurons derived from global Ube3B knockout mice found alterations in two presynaptic markers. Numbers of Synapsin puncta were reduced, while inhibitory presynaptic terminals marked with VGAT were increased. Subsequent studies using conditional Ube3B knockout mice examined the effects of Ube3B in neurons (Ambrozkiewicz et al., 2021). Electrophysiological results with cultured excitatory hippocampal neurons from Ube3B conditional knockouts suggested presynaptic transmission is potentially increased. Reduced presynaptic markers in global knockout neurons and increased presynaptic excitatory transmission in neurons from conditional knockouts might seem contradictory at first pass. However, this could simply reflect differing genetic strategies, the use of autaptic culture for one study, or differences between hippocampal and cortical neurons. Regardless, what emerges consistently from both studies is that Ube3B could affect presynaptic function and/or formation. These findings also suggest there could be an imbalance in presynaptic connections formed by excitatory and inhibitory neurons in the absence of Ube3B. Whether this is the case or not will require a more thorough examination of presynaptic development and function, as well as future in vivo presynaptic studies. While the BCKDK kinase and the protein phosphatase subunit Ppp3cc are Ube3B substrates (Cheon et al., 2019; Ambrozkiewicz et al., 2021), it is unclear if Ube3B functions through these targets or others to influence presynaptic development or transmission (Fig 2C). Whether Ube3B regulates different substrates to influence pre- versus postsynaptic function remains an interesting but untested possibility.

Thin, HERC1 and Ariadne-1 in presynaptic transmission and synaptic vesicle dynamics

Multiple ubiquitin ligases have been shown to affect presynaptic plasticity and transmission as well as synaptic vesicle dynamics in different models. Our initial discussion will center on ubiquitin ligases that influence excitatory presynaptic transmission. We will comment on ligases that affect inhibitory presynaptic transmission later in our discussion.

One widely used model to study presynaptic glutamatergic transmission is the NMJ of Drosophila (Menon et al., 2013; Chou et al., 2020). The fly NMJ is also a preeminent model for evaluating presynaptic homeostatic plasticity, synaptic scaling that occurs in response to excess excitatory transmission (Davis & Muller, 2015). The ubiquitin-proteasome system (UPS) was initially implicated in NMJ synaptic transmission and homeostatic plasticity by pharmacologically and genetically impairing the proteasome (Speese et al., 2003; Wentzel et al., 2018). The genetic tractability of the fly model subsequently facilitated a large-scale, candidate screen that sought to identify a ubiquitin ligase that provides specificity for UPS effects on homeostatic plasticity (Baccino-Calace et al., 2022). This unbiased approach revealed the RING ubiquitin ligase Thin/TRIM32 as a key inhibitor of homeostatic plasticity and synaptic vesicle release. Genetic interaction studies indicated that Thin inhibits Dysbindin to restrict the readily releasable pool (RRP) of synaptic vesicles (Fig 3A). While one mechanism for how Thin regulates presynaptic release is in hand, whether Thin acts through Dysbindin to affect presynaptic homeostatic plasticity awaits further studies. The importance of expanded studies on Thin is highlighted by emerging links between its human ortholog TRIM32 and brain disorders, such as autism and attention deficit hyperactivity disorder (Lionel et al., 2011; Lionel et al., 2014).

Fig 3. Thin, HERC1 and Ariadne-1 in presynaptic excitatory transmission, synaptic vesicle dynamics and homeostatic plasticity.

A) At the Drosophila NMJ, Thin/TRIM32 inhibits Dysbindin to regulate the ready-releasable pool (RRP) and synaptic vesicle release. Thin also regulates presynaptic homeostatic plasticity via unknown downstream mechanisms. B) At mammalian central synapses and NMJ, HERC1 promotes the RRP and synaptic transmission. At central synapses, HERC1 promotes SV2A and V-Glut accumulation. C) Ari-1/ARIH1 monoubiquitinates Comt/NSF to promote synaptic vesicle release at the Drosophila NMJ.

Now let us turn to ubiquitin ligases found to affect presynaptic terminals at mammalian central synapses. A recent study using cultured hippocampal neurons found that HERC1, a HECT ligase, regulates presynaptic terminals (Montes-Fernandez et al., 2020). HERC1 function was altered by deriving hippocampal neurons from a HERC1 mutant mouse, tambaleante (Tbl). EM and light microscopy studies indicate that synaptic vesicle numbers, synaptic vesicle markers, and active zone numbers are all reduced at presynaptic terminals of Tbl mutant neurons. Labeling synaptic vesicles with FM1–43 dye via endocytosis showed that the RRP and reserve vesicle pool are both reduced in Tbl mutant neurons, which suggests a presynaptic release capacity is reduced. These observations in Tbl mutant hippocampal neurons are consistent with two prior findings. First, the amygdala of Tbl mutant mice have reductions in the SV2A synaptic vesicle protein and V-Glut (Perez-Villegas et al., 2018). Second, HERC1 regulates synaptic transmission at the mammalian NMJ by influencing the RRP (Bachiller et al., 2015) (Fig 3B). These findings indicate that HERC1 influences presynaptic composition and vesicle release, which could contribute to memory defects observed in Tbl mutant mice (Perez-Villegas et al., 2018). However, we note an important caveat – how HERC1 function is altered in the Tbl mutant mouse remains unclear. Thus, experiments with cleaner more targeted genetic tools will be necessary to determine whether HERC1 is a positive or negative regulator of presynaptic transmission. Increasing the HERC1 genetic toolkit could be instrumental in better understanding the pathology of a neurodevelopmental disorder associated with HERC1 (Ortega-Recalde et al., 2015; Nguyen et al., 2016; Perez-Villegas et al., 2022).

A third ubiquitin ligase recently shown to affect presynaptic release, Ariadne-1 (Ari-1), was also revealed by work in Drosophila (Ramirez et al., 2021). Ari-1 is a conserved RBR ligase that is orthologous to mammalian Ariadne RBR ubiquitin protein ligase 1 (ARIH1) (Aguilera et al., 2000). Proteomics from Drosophila brain revealed numerous putative Ari-1 substrates including Comt which is the fly ortholog of NSF, a component of the synaptic vesicle release machinery (Ramirez et al., 2021). Biochemistry from flies and cell-based studies with human proteins confirmed NSF/Comt as a conserved, mono-ubiquitinated substrate for Ari-1/ARIH1. Consistent with Ari-1 regulating Comt/NSF, genetic studies with multiple Ari-1 mutant alleles demonstrated reduced presynaptic transmission at the fly NMJ. Results from complex genetic interaction studies using a whole-animal behavioral readout suggested that Ari-1 and Comt function together to regulate locomotor recovery following heat shock. These findings indicate that Ari-1 regulates presynaptic transmission, and suggest that Ari-1 functions with Comt/NSF to regulate synaptic transmission and NMJ function (Fig 3C).

We look forward to future studies aimed at better understanding how Thin/TRIM32, HERC1 and Ari-1/ARIH1 shape excitatory presynaptic transmission. The present state of the field raises two questions for consideration. 1) Do Thin/TRIM32, HERC1 and Ari-1/ARIH1 represent species or neuron-specific mechanisms for regulating presynaptic transmission? 2) Alternatively, are these ligases working coordinately in a ubiquitin ligase network to modulate presynaptic formation and transmission in excitatory neurons? Given that the ligases Staring and Scrapper further regulate presynaptic transmission (Chin et al., 2002; Yao et al., 2007), an extensive ubiquitin ligase network is potentially required to optimally tune presynaptic, excitatory transmission.

Nedd4 in presynaptic metabotropic glutamate receptor trafficking

Nedd4 is a HECT family ubiquitin ligase that has broad functions within and outside the nervous system (Rotin & Kumar, 2009; Rotin & Prag, 2024). Prior studies have shown Nedd4 functions postsynaptically to regulate synapse formation at the neuromuscular junction of flies and mice. This occurs via Nedd4-mediated degradation of GPCRs such as AMPA-type glutamate receptors and the sorting Nexin SH3PX1 (Ing et al., 2007; Liu et al., 2009; Zhu et al., 2017; Wasserman et al., 2019).

A recent study using cell-based approaches and primary neurons has identified a further role for Nedd4 in presynaptic biology (Lee et al., 2019). Metabotropic glutamate receptor 7 (mGlu7) is a presynaptic neuromodulatory GPCR that responds to glutamate release by stimulating inhibitory G protein signaling. Nedd4 was found to ubiquitinate and degrade agonist stimulated mGlu7 in cell-based experiments and cortical neurons. Interestingly, β-arrestin functions as a binding protein and adaptor that recruits Nedd4 to mGlu7, and Nedd4 targets mGlu7 for degradation via both proteasomal and lysosomal mechanisms (Fig 4). It is possible that the Nedd4/β-arrestin complex also sterically inhibits G-protein binding to mGlu7, but this remains untested. Additionally, binding of the β-arrestin/Nedd4 complex was shown to be required for mGlu7-dependent ERK signaling. With Nedd4 identified as an inhibitor of mGlu7, it will be valuable to determine how broad its regulator effects are on mGlu7 in different types of excitatory circuits and to further establish how its effects on mGlu7 shape glutamatergic transmission in the central nervous system.

Fig. 4. Nedd4 in presynaptic metabotropic glutamate receptor trafficking.

Summary of results from cell-based studies and primary rat cortical neurons which showed that Nedd4 ubiquitinates mGlu7 leading to both proteasomal and lysosomal degradation. β-arrestin binds and recruits Nedd4 to mGlu7. Nedd4/β-arrestin binding is required for mGlu7-dependent ERK signaling.

MYCBP2/RPM-1 influences acetyltransferases, microtubule stability and NMNAT

Next, we turn to the PHR group of RING ubiquitin ligases, which are conserved regulators of axon termination and synaptogenesis that also play important post-developmental roles in axon degeneration (Grill et al., 2016; Virdee, 2022). The PHR ligases derive their name from human, Drosophila, and C. elegans orthologs: PAM (also called MYCBP2), Highwire (Hiw), and RPM-1, respectively. In rodents, MYCBP2 is also referred to as Phr1. Recent structural and biochemical studies have shown that MYCBP2 has an unusual RING-cys-relay (RCR) enzymatic mechanism, forms an atypical Skp/Cullin/Fbox (SCF) complex that lacks a Cullin, and can ubiquitinate threonine and serine residues while canonical ubiquitination occurs at lysine (Saiga et al., 2009; Desbois et al., 2018; Pao et al., 2018). PHR proteins form a complex with the F-box protein FBXO45/FSN-1, which mediates substrate recognition (Liao et al., 2004; Saiga et al., 2009; Sharma et al., 2014; Desbois et al., 2018). Prior work from C. elegans, has demonstrated that PHR proteins are enormous ubiquitin ligase signaling hubs that utilize both enzymatic and non-enzymatic mechanisms to regulate an extensive signaling network (Grill et al., 2016). Here, we comment only on recent studies that have expanded our understanding of the MYCBP2 signaling network, cellular mechanisms controlling presynaptic development and function, and the recent discovery of a childhood neurodevelopmental spectrum disorder associated with MYCBP2.

From a clinical and translational perspective, the importance of the MYCBP2 ubiquitin ligase signaling hub has grown substantially with the recent discovery of human genetic variants in MYCBP2 that cause a neurodevelopmental disorder, referred to as MYCBP2-related developmental delay with corpus callosum defects (MDCD) (AlAbdi et al., 2023). Multiple patients with de novo variants in MYCBP2 were independently identified, and found to display a spectrum of neurobehavioral abnormalities including developmental delay, intellectual disability, autism and seizure. Neurobehavioral abnormalities are often accompanied by anatomical defects in the corpus callosum. To examine pathogenicity, patient-associated variants were CRISPR edited into C. elegans RPM-1/MYCBP2 and functional effects were evaluated in vivo. Outcomes for axon development and behavioral habituation to repeated mechanosensory stimulation demonstrated that patient-associated missense variants result in variable loss of function. Patient-associated variants also resulted in axonal accumulation of the autophagosome marker LGG-1/LC3, which is consistent with the previous finding that RPM-1 inhibits autophagy.

RPM-1 regulates neuronal autophagy in C. elegans by ubiquitinating and degrading the autophagy initiating kinase UNC-51/ULK (Crawley et al., 2019). This finding initially emerged from a proteomics screen for RPM-1 ubiquitination substrates. Increased levels of endogenous UNC-51 in rpm-1 mutants and cell-based biochemical results further validated UNC-51/ULK as a conserved RPM-1/MYCBP2 substrate that is polyubiquitinated and degraded by the proteasome. In the C. elegans nervous system, RPM-1 inhibits UNC-51 to restrict axonal autophagosome formation. RPM-1 inhibition of UNC-51 promotes axon termination and glutamatergic synapse maintenance in mechanosensory neurons. Consistent with this, RPM-1 inhibition of UNC-51 also affects behavioral habituation to repeated mechanosensory stimulation. Amongst ubiquitin ligases that influence presynaptic or axonal autophagy, RPM-1 is the sole inhibitor of autophagy as SIAH1, Parkin, and Atg5 all promote presynaptic autophagy (Fig 5A; 7). Given the importance of autophagy in both development and degenerative disease states, it will be valuable to understand whether this emerging network of ligases functionally integrates to yield proper levels, localization, or developmental timing for presynaptic and axonal autophagy.

Fig 5. MYCBP2 influences autophagy, acetyltransferases, microtubule stability, and NMNAT.

Previous studies showed that the MYCBP2 ubiquitin ligase (Phr1 in mice) functions as a ubiquitin ligase signaling hub to regulate multiple events in neuronal development. Here, we highlight recently discovered functions and substrates for MYCBP2 in C. elegans, Drosophila and vertebrates. A) In C. elegans, the MYCBP2 ortholog RPM-1 was recently shown to promote synapse maintenance by positively regulating two microtubule stabilizing proteins, the acetyl transferase ATAT-2 and the microtubule binding protein PTRN-1. RPM-1 polyubiquitinates and inhibits the autophagy initiating kinase UNC-51/ULK to restrict autophagy and influence both synapse maintenance and axon termination. RPM-1 regulation of axon termination also occurs via ubiquitination and inhibition of the atypical cyclin dependent kinase CDK-5, as well as positive genetic interactions with the VAB-1 Ephrin receptor. B) At the fly NMJ, the MYCBP2 ortholog Hiw promotes axon terminal pruning and polyubiquitinates NMNAT to inhibit presynaptic transmission. C) MYCBP2 binds and stabilizes the EPHB2 receptor to regulate EPHB2-mediated axon repulsion.

Fig. 7. A network of ubiquitin ligases controls several presynaptic processes to influence synapse development, maintenance, and function.

Diagram showing all discussed ubiquitin ligases and their functional roles at the presynaptic terminal. Each ligase has a unique color to more easily trace associated arrows. Processes affected include synaptic transmission, modulation, synaptic vesicle dynamics, endosomal sorting (retromer), autophagy, microtubule stability, mitochondrial regulation, axon termination, and pruning. Please note Scrapper, Staring and APC ligases are not discussed here, but also have presynaptic roles. Thus, the presynaptic ubiquitin ligase network is potentially even more extensive than depicted here.

Proteomics and CRISPR-based biochemistry with native RPM-1 further revealed another kinase, CDK-5, as an RPM-1 substrate (Desbois et al., 2022). Cdk5 is an atypical cyclin-dependent kinase that is highly conserved with functions in neuronal development and synaptic plasticity. Aberrant Cdk5 function is linked to neurodegenerative diseases, including Alzheimer’s Disease and Parkinson’s Disease. Genetic results indicated that RPM-1 promotes axon termination in mechanosensory neurons in part by inhibiting CDK-5. RPM-1 has differing functional relationships with CDK-5 at distinct subcellular locations, as RPM-1 functions coordinately with CDK-5 to regulate synapse formation (Fig 5A). These studies highlight an important point – ubiquitin ligases can have layers of functional interactions with a substrate within different compartments of the same neuron.

Another cellular process that RPM-1 influences to regulate presynaptic development is microtubule dynamics (Borgen et al., 2019). Genetic and pharmacological findings with C. elegans mechanosensory neurons indicate that RPM-1 regulates presynaptic maintenance via ATAT-2 tubulin acetyltransferase activity, which promotes microtubule stability. Consistent with these results, ATAT-2 regulated responses to behavioral habituation to repeated mechanosensory stimulation. Multi-gene interaction studies demonstrated that RPM-1 is an upstream, positive regulator of ATAT-2 in the context of synapse maintenance and behavioral habituation (Fig 5A). While RPM-1 promotes microtubule stability to facilitate presynaptic maintenance, it acts as a microtubule destabilizer that regulates growth cone collapse and axon termination at a physically distinct location within the same mechanosensory neurons (Borgen et al., 2017b)

Studies on Drosophila Hiw using two different types of neurons have substantially expanded the field’s appreciation for PHR protein effects on axon pruning and presynaptic transmission. In the central nervous system (CNS), Hiw was found to cell-autonomously promote axon pruning at giant fiber termination sites where presynaptic terminals are formed (Borgen et al., 2017a). Surprisingly, Hiw can act non-cell autonomously in neighboring glia to promote synapse formation. In the peripheral nervous system of Drosophila and C. elegans, Hiw and RPM-1 function presynaptically at the NMJ and in mechanosensory neurons to regulate synapse formation through ubiquitination and degradation of the DLK/Wallenda MAPK pathway (Nakata et al., 2005; Collins et al., 2006; Yan et al., 2009). Hiw mutants have not only synaptic overgrowth defects at the NMJ, but also reduced presynaptic transmission. While synapse formation phenotypes are mediated by DLK signaling, synaptic transmission defects do not occur via this pathway. Electrophysiological and genetic interaction studies have now revealed inhibition of the NAD-synthase enzyme Nmnat as the mechanism behind Hiw effects on transmission (Fig 5B) (Russo et al., 2019). Previous studies established Nmnat as a MYCBP2/Hiw ubiquitination substrate that is degraded by the proteasome (Babetto et al., 2013; Yang et al., 2015; Desbois et al., 2018). Nmnat colocalizes at presynaptic active zones with Bruchpilot/ERC/Elks, and Nmnat overexpression results in decreased presynaptic neurotransmitter release (Russo et al., 2019). While Hiw was known to inhibit Nmnat to mediate axon degeneration in Drosophila (Xiong et al., 2012), these new findings establish a key physiological relationship between Hiw and Nmnat in presynaptic transmission. Thus, Hiw functions through different presynaptic signaling mechanisms to regulate presynaptic formation and transmission at the fly NMJ.

At present, much less is known about how human MYCBP2 and its mouse ortholog Phr1 shape presynaptic development and function compared to invertebrates. Prior studies have shown that Phr1 is a conserved regulator of synapse formation at the mouse NMJ similar to findings from Drosophila and C. elegans (Burgess et al., 2004; Bloom et al., 2007). Phr1 also has prominent roles in axon degeneration as we noted above and in axon development via effects on DLK and TSC/mTOR signaling (Lewcock et al., 2007; Han et al., 2012; Grill et al., 2016). A recent study has now demonstrated that along with degrading and restricting DLK and TSC, MYCBP2 can have opposing positive regulatory effects on EPHB2 Ephrin receptor stability and signaling (Chang et al., 2024). Proteomic and biochemical findings demonstrated that human MYCBP2 and its F-box protein FBXO45 bind and promote stabilization of EPH2B (Fig 5C). Consistent with these findings, impairing MYCBP2 reduces EPHB2-mediated axon repulsion and signaling in cell-based assays and neuronal explants from chick. Given these findings and the prominent role that MYCBP2 and EphB2 play in synapse formation, we will be intrigued to see whether future vertebrate studies determine if a MYCBP2/Eph2B signaling axis regulates presynaptic formation.

HUWE1 and LNX1/2 regulate GABAergic and Glycinergic inhibitory neurons

In recent years, there has been important progress in identifying presynaptic ubiquitin ligases that affect the function of inhibitory neurons. In general, we still know relatively little about presynaptic ligases that affect inhibitory transmission compared to those influencing excitatory transmission. At present, presynaptic ligases affecting inhibitory transmission break down into two groups based on the category of inhibitory neuron they regulate, GABAergic or glycinergic.

With regard to glycinergic neuron function, two ubiquitin ligases in the RING family have emerged as potential regulatory players, LNX1 and LNX2 (de la Rocha-Munoz et al., 2019). GlyT2 is a neuronal transporter that allows reuptake of glycine and packaging into synaptic vesicles. Interestingly, patient-associated variants in GlyT2 lead to a childhood disorder called hyperekplexia (Rees et al., 2006; Harvey et al., 2008). Results from cell-based assays showed that LNX1 and LNX2 are sufficient to bind, polyubiquitinate and regulate levels of GlyT2 (Fig 6A). In hippocampal neurons, knockdown of LNX2 resulted in increased levels of GlyT2 and increased glycine transport rates. These findings indicate that LNX is capable of inhibiting GlyT2 thereby affecting glycine reuptake in neurons. Whether LNX1/2 affects glycinergic neuron function or transmission in cultured neurons or possibly in vivo awaits further experiments.

Fig. 6. LNX1/2 and HUWE1 regulate presynaptic function in Glycinergic and GABAergic inhibitory neurons.

A) In hippocampal neurons, LNX1/2 polyubiquitinates GlyT2 resulting in its endocytosis and degradation thereby reducing glycine reuptake. LNX1/2 can ubiquitinate active zone components such as Liprin-α and CAST/ERC, but how this influences presynaptic development or transmission remains unclear. B) At the C. elegans inhibitory NMJ, HUWE1/EEL-1 forms a complex with OGT-1 with both players functioning coordinately to regulate synaptic transmission in GABAergic motor neurons.

Earlier studies showed LNX1/2 also binds to presynaptic active zone components such as Liprin-alpha and CAST/ERC2 (Higa et al., 2007; Lenihan et al., 2017b), and can colocalize with CAST at presynaptic terminals of cultured hippocampal neurons (Fig 6A) (Higa et al., 2007). Interestingly, cell-based proteomic and biochemical studies indicate that LNX1 binds and ubiquitinates Liprin-alpha and CAST/ERC2 (Lenihan et al., 2017b). These proteomic studies also hinted at the possibility that LNX1/2 could form a complex with other presynaptic ligases such as MYCBP2. At present, it remains unclear whether LNX1 and 2 ubiquitination of active zone components and GlyT2 affects neuronal function and transmission. However, LNX1/2 double knockout mice display reduced anxiety-like behaviors and sex-variable alterations in locomotor behavior suggesting potential changes in neuronal activity or presynaptic transmission could be occurring (Lenihan et al., 2017a).

We now turn to HUWE1, which is an extremely large, conserved HECT family ubiquitin ligase that regulates GABAergic transmission in C. elegans (Giles & Grill, 2020). Interestingly, several independent clinical studies have identified multiple missense variants in HUWE1 that are associated with a neurodevelopmental spectrum disorder we will refer to as HUWE1-associated neurodevelopmental disorder (HANDD) (Friez et al., 2016; Moortgat et al., 2017; Giles & Grill, 2020). Patients with HANDD feature intellectual disability as a primary neurobehavioral abnormality with further reports of autism and seizure. To date, studies showing HUWE1 affects GABAergic neurons have only been performed in C. elegans which has a single, highly conserved ortholog called EEL-1 (Opperman et al., 2017). While HUWE1/EEL-1 is broadly expressed in the C. elegans nervous system, it was found to preferentially affect presynaptic transmission in the inhibitory GABAergic neurons (Fig 6B). This was assessed using both electrophysiological and pharmacological readouts of GABA neuron function. Consistent with these findings, EEL-1 localizes to GABAergic presynaptic terminals. Interestingly, genetic outcomes with transgenic rescue experiments in C. elegans suggest two patient-associated HUWE1 variants reduce EEL-1/HUWE1 gene function (Opperman et al., 2017). Notably, the role of EEL-1 in the nervous system is not limited solely to GABAergic transmission, as genetic results indicate EEL-1 can function as part of the RPM-1 signaling network to regulate presynaptic development in both inhibitory GABAergic and excitatory cholinergic motor neurons (Opperman et al., 2017).

More recent findings have begun to reveal the mechanisms that EEL-1 utilizes to regulate presynaptic GABAergic function (Giles et al., 2019). The first is enzymatic HECT ubiquitin ligase activity, which was tested using transgenic rescue experiments with EEL-1 point mutations that reduce enzymatic ligase activity. The second mechanism involves the OGT-1 glycosyltransferase, which was shown to be a conserved binding protein for C. elegans EEL-1 and human HUWE1 via proteomics and biochemistry. Like EEL-1, OGT-1 localizes to GABAergic presynaptic terminals in C. elegans. EEL-1 and OGT-1 regulate GABA neuron function by forming a protein complex and functioning coordinately to regulate GABA neuron function (Fig 6B). Genetic results indicate that OGT-1 cannot be ubiquitinated and degraded by EEL-1. This suggests that HUWE1 also regulates GABA neuron function via presently unknown substrates. OGT-1 was found to function independent of its catalytic glycosyltransferase activity to regulate GABA neuron function. Interestingly, these findings could have clinical importance, as patients with neurobehavioral abnormalities and genetic variants in OGT-1 have been identified, similar to HUWE1/EEL-1 (Vaidyanathan et al., 2017; Willems et al., 2017; Giles et al., 2019). Thus, the HUWE1/OGT-1 signaling axis is relevant to both GABA neuron function and neurodevelopmental disorders. Collectively, these studies indicate that HUWE1 utilizes both enzymatic and non-enzymatic mechanisms to regulate presynaptic function in GABA neurons. Whether HUWE1 and HANDD-associated variants affect inhibitory or excitatory neuronal function in mammals remains an open and important question.

Conclusion

Recent research has revealed a multiple component ubiquitin ligase network that regulates many facets of presynaptic development and function (Fig 7). This includes ligases that affect synaptic transmission, synaptic vesicle dynamics, endosomal sorting, autophagy, neuromodulation, microtubule stability, presynaptic maintenance, axon termination and synaptic pruning. Importantly, this presynaptic ligase network functions through sophisticated mechanisms. While several ligases directly target synaptic proteins for degradation, there are also examples where ligases regulate more extensive downstream signaling pathways to control neuronal processes. Furthermore, ligases function through polyubiquitination or monoubiquitination of substrates to degrade or modulate their activity, respectively. While the presynaptic ligase network affects many cellular processes, we see multiple core themes emerging. 1) Ubiquitin ligases are key regulators of neuronal autophagy with different ligases promoting and inhibiting autophagy. Thus, ubiquitin ligases are likely to be critical signaling players mediating precise, balanced levels of presynaptic autophagy. 2) Synaptic transmission is regulated by several ubiquitin ligases that utilize different molecular mechanisms. Understanding how each ligase affects transmission and in which neurons remains an important future direction. 3) Extensive clinical genetic studies now indicate that presynaptic ubiquitin ligases form a core group of players with prominent roles in several genetically distinct neurodevelopmental disorders. This builds substantially upon prior links between presynaptic ligases and degenerative diseases. As a result, presynaptic ubiquitin ligases are potentially suitable molecular targets for future therapies aimed at managing neurological conditions that afflict both the developing and aged nervous system.