In recent times, our knowledge of the roles ubiquitin plays in multiple cellular processes has expanded exponentially, with one example being the role of ubiquitin in receptor endocytosis and trafficking. This has prompted a multitude of studies examining how the different machinery involved in the addition and removal of ubiquitin can influence this process. Multiple deubiquitylating enzymes (DUBs) have been implicated either in facilitating receptor endocytosis and lysosomal degradation or in rescuing receptor levels by preventing endocytosis and/or promoting recycling to the plasma membrane. In this review, we will discuss in detail what is currently known about the role of DUBs in regulating the endocytosis of various transmembrane receptors and ion channels. We will also expand upon the role DUBs play in receptor sorting at the multivesicular body to determine whether a receptor is recycled or trafficked to the lysosome for degradation. Finally, we will briefly discuss how the DUBs implicated in these processes may contribute to the pathogenesis of a range of diseases, and thus the potential these have as therapeutic targets.
In the last decade, our knowledge of the different roles ubiquitin plays has rapidly expanded and it is now apparent that ubiquitin has a major impact on nearly every complex cellular process in some way. Ubiquitin is a highly conserved 8 kDa protein, which was initially implicated in proteasome-mediated degradation, but has more recently been shown to play multiple roles when covalently attached to substrate proteins [1–3].
Ubiquitylation (ubiquitination) is a post-translational modification that involves covalent attachment of ubiquitin to target proteins. Ubiquitin is mostly conjugated via its C-terminus to the ε-amino group of lysine residues on target proteins, or other ubiquitin molecules, by the action of an enzymatic cascade involving E1 ubiquitin-activating enzymes, E2 ubiquitin-conjugating enzymes, and E3 ubiquitin ligases . A substrate protein can have a single ubiquitin residue attached (monoubiquitylation), have multiple single ubiquitins attached (multi-monoubiquitylation), or be modified with multiple ubiquitin residues in the form of a polyubiquitin chain. These chains are formed by attaching multiple ubiquitins to one another via their N-termini (linear ubiquitin), or one of the seven lysine (K) residues (K6, K11, K27, K29, K33, K48, and K63) present in each ubiquitin . Polyubiquitin chains vary in length and adopt a range of different topologies, depending on the residue/s used to attach them together, which result in different functional outcomes for the target protein, all of which has been comprehensively reviewed before .
Deubiquitylating (deubiquitinating) enzymes (DUBs) reverse the ubiquitylation of target proteins in a role similar to that of phosphatases in kinase/phosphatase regulatory pathways . DUBs are proteases that remove ubiquitin to reverse the modification of target proteins, or remodel polyubiquitin chains , and therefore often act as the final arbiters of protein fate . DUBs must also identify their targets among a vast array of different protein–ubiquitin and ubiquitin–ubiquitin linkages in different subcellular and functional contexts . More than 90 potential DUBs have been identified in mammals and they are divided into five main classes dependent on distinct characteristics. These are ubiquitin C-terminal hydrolases (UCHs: 4 members), ubiquitin-specific proteases (USPs/UBPs: 58 members), ovarian tumour proteases (OTUs: 14 members), JAMM (Jab1/Pab1/MPN metalloenzyme; 12 members) motif proteases, and Machado-Joseph disease protein domain proteases (MJDs: 4 members) [6,9,10]. UCH, USP, OTU, and MJD DUBs are all thiol proteases with an active-site cysteine residue that acts as a nucleophile to facilitate the attack on the lysine–glycine isopeptide linkage of ubiquitylated proteins. JAMM proteases alternatively utilise a zinc metalloprotease domain to cleave the bonds between ubiquitin and the target protein . A sixth DUB family has been proposed based on the characterisation of a new domain in monocyte chemotactic protein-induced protein 1 (MCPIP1), which displays deubiquitinating activity. MCPIP1 contains cysteine boxes, characteristic of cysteine proteases, and a CCCH-type zinc-finger domain, both of which are required for its ability to cleave ubiquitin . However, a recent study indicates MCPIP1 complexes with USP10 and this may be responsible for the proposed MCPIP1 DUB activity .
DUBs have several functions in the ubiquitin pathway, including activation of ubiquitin pro-proteins, ubiquitin recycling, and removal of ubiquitin from target proteins . Ubiquitin is always expressed as a pro-protein either fused to ribosomal proteins or as a linear polyubiquitin chain consisting of multiple ubiquitin monomers. Therefore, it needs processing into the monomer prior to being used in the ubiquitin system. However, DUBs are most widely associated with the removal of ubiquitin monomers and polyubiquitin chains from target proteins to affect their fate by rescuing them from degradation (proteasomal or lysosomal) [15,16], modulating their activity , or altering their binding partners . DUBs have also been associated with the regulation of multiple cellular functions, including gene expression , cell-cycle progression , intracellular signalling , DNA repair [21,22], and chromatin remodelling . However, a particular area where the influence of DUBs has grown over the past few years is in the regulation of receptor endocytosis and trafficking, and thus we will review that here.
Receptor endocytosis and ubiquitin
A multitude of transmembrane receptors and ion channels are found at the plasma membrane, and their activity needs to be tightly regulated to allow their proper function. One mechanism of regulation is modulating their availability at the plasma membrane through their rapid removal via endocytosis . Endocytosis involves the internalisation of a portion of the plasma membrane, incorporating the desired receptor/channel, into internal membrane compartments. This is required for multiple biological processes, including cell motility, intracellular signalling, antigen presentation, and synaptic vesicle (SV) recycling [24–26]. Diverse endocytosis mechanisms have evolved to enable delivery of cargo to specific intracellular compartments and these pathways are classified into several distinct types depending on morphology, lipid, accessory components, coat components, and specific cargoes involved . These distinct routes of endocytosis include clathrin-mediated endocytosis (CME), caveolar-type endocytosis, clathrin-independent carriers/GPI-enriched early endosomal compartments, flotillin-dependent endocytosis, phagocytosis, macropinocytosis, and circular dorsal ruffles, all of which are extensively reviewed elsewhere . Endocytosis also allows many receptors to interact with relevant signalling molecules and causes the cell to respond to particular extracellular cues. The fate of the receptor upon endocytosis (recycling or degradation) also regulates the sensitivity of the cell to a specific ligand , as well as coordinating receptor distribution and influencing downstream signalling . Therefore, endocytosis and degradation of receptors/channels needs to be tightly controlled to avoid aberrant and potentially pathological signalling.
Ubiquitin plays a fundamental role in endocytosis, with studies in yeast indicating that many receptors require monoubiquitylation to facilitate internalisation and trafficking to the vacuole for degradation [30,31]. In mammals, the role of ubiquitin is more complex with many receptors being ubiquitylated upon interacting with their ligand and this ubiquitylation then triggers their endocytosis and trafficking to the lysosome for degradation . However, while monoubiquitylation is often sufficient to trigger internalisation and lysosomal targeting, many receptors require the addition of K63-linked polyubiquitin chains to obtain maximal rates of endocytosis [32–34]. Upon endocytosis, receptors are internalised into early endosomes , which act as signalling platforms by coupling these activated receptors with specific signalling machinery. The early endosomes can also serve as a central sorting terminal [36,37] with the receptors either being recycled back to the plasma membrane or directed to the lysosome for degradation via late endosomes and multivesicular bodies (MVBs). Ubiquitin-dependent receptor degradation requires sorting of the receptors into vesicles that invaginate into the interior of MVBs and facilitate the trafficking of these receptors to the lysosome, although ubiquitin must be removed to allow their incorporation into the vesicles of the MVBs. MVBs fuse with the lysosome and the vesicles, including the incorporated receptors, are degraded [1,38,39].
The identification of aberrantly activated receptor tyrosine kinases, such as epidermal growth factor receptor (EGFR), in cancer and other diseases [40,41] has led to a recent focus on the mechanisms by which these receptors are regulated. This has driven forward our understanding of endocytosis considerably, and the prominent role of ubiquitin in facilitating receptor endocytosis has prompted extensive studies examining how the ubiquitylation machinery, particularly DUBs, influences this process.
DUBs and endocytosis
DUBs and EGFR endocytosis
EGFR is the most studied receptor with regard to endocytosis, and it is almost exclusively internalised via CME when stimulated with a low dose of its ligand EGF [42,43]. However, at higher EGF concentrations, a substantial fraction of EGFR is internalised via clathrin-independent routes prior to trafficking to the lysosome for destruction [42,44]. Recent studies suggest that while EGFR is monoubiquitylated during CME, the switch to clathrin-independent routes of endocytosis correlates with the stable binding of the E3 ligase Cbl, resulting in EGFR polyubiquitylation . Many DUBs have been implicated in the control of EGFR endocytosis (Figure 1) and a recent study, which did an siRNA screen of 92 DUBs to determine their impact on EGFR endocytosis, suggests that several others may also be involved .
DUB regulation of EGFR endocytosis.
USP9X (Fam) blocks EGFR endocytosis and trafficking to the lysosome for degradation . Initially, this was attributed to its action on the E3 ligase Itch that ubiquitylates Cbl and triggers its degradation via the proteasome [48,49]. USP9X was shown to rescue Itch from degradation, trigger Cbl degradation, and block EGFR polyubiquitylation and lysosomal trafficking . However, a more recent study indicates that USP9X does not affect EGFR ubiquitylation, but instead deubiquitylates Eps15 (epidermal growth factor receptor pathway substrate 15), an adaptor required for endocytosis . This study also shows that USP9X depletion impedes EGFR lysosomal targeting and degradation after it enters the cell, suggesting that USP9X has an additional role post-internalisation . This fits with its localisation to internal vesicles, but further investigation will be required to fully explain how this role is mediated .
Cezanne-1 and Cezanne-2, members of the OTU family, were identified as negative regulators of EGFR degradation in an siRNA screen. Cezanne-1 interacts with EGFR, stabilises it, and enhances its signalling. This was explained by the observation that Cezanne-1 deubiquitylates EGFR and prevents its lysosomal degradation . However, Cezanne-1 was also shown to interact with the E3 ligases Itch and Nedd4, which regulate EGFR endocytosis, and this could play a role in this regulation, although it was not investigated in the study .
Like Cezanne-1, USP2a (USP2-69) binds to and deubiquitylates EGFR, stabilises it, and amplifies its signalling. In particular, USP2a increases EGFR recycling and the proportion of EGFR at the plasma membrane. Indeed, USP2a, at least partially, localises to early endosomes, indicating that it deubiquitylates EGFR post-internalisation and directs it for recycling to the plasma membrane preventing its degradation .
EGFR activation induces USP17 expression and this was initially believed to regulate EGFR downstream signalling . However, subsequent studies have shown that USP17 is actually required for EGFR CME, due to its necessity for the production of phosphoinositol-4,5-phosphate (PIP2), a lipid which anchors many CME adaptors, at the plasma membrane . USP17 depletion mislocalises phosphoinositol-5-phosphate kinase β, an enzyme known to produce PIP2, and previously shown to be required for transferrin receptor (TfR) and EGFR endocytosis [55,56]. The role of USP17 in CME was further supported by the observation that it is required for endocytosis of TfR, an archetypal substrate for CME . USP17's role is probably more widespread as its expression is induced by multiple stimuli [interleukin (IL)-4, IL-6, IL-8, and stromal cell-derived factor 1] [53,57], indicating a potential role in the endocytosis of multiple receptors.
DUBs and endocytosis of other transmembrane receptors and ligands
Several of the DUBs implicated in EGFR endocytosis also influence the endocytosis of other receptors (Figure 2). In particular, fat facets, the Drosphilia homologue of USP9X, was initially shown to interact with liquid facets (epsin homologue) and to deubiquitylate it causing endocytosis of the Notch ligand Delta [58,59]. USP9X has subsequently been shown to interact with and deubiquitylate epsin 1, suggesting that this is a conserved action . The E3 ligase Mind bomb is also required for Delta endocytosis and signalling through Notch. Mind bomb auto-ubiquitylates and is again degraded by the proteasome unless deubiquitylated by USP9X . Therefore, as with EGFR, USP9X may have multiple roles in the regulation of Δ endocytosis
DUB regulation of receptor endocytosis.
USP2 also regulates low-density lipoprotein receptor (LDLR) endocytosis. The E3 ligase inducible degrader of the LDLR (IDOL) ubiquitylates LDLR and targets it for clathrin-independent endocytosis and lysosomal degradation . Both isoforms of USP2 (USP2a, USP2b [USP2-45]) interact with IDOL, deubiquitylate it, and stabilise it by blocking its proteasomal degradation. However, rather than enhancing LDLR degradation, this USP2, IDOL, and LDLR complex blocks the action of IDOL on LDLR and leads to LDLR stabilisation . It is not clear yet how USP2 can stabilise IDOL but block its action upon LDLR, and further work will be required to decipher the underlying mechanism.
DUB-2A, a murine orthologue of USP17, reduces the ubiquitylation and lysosomal degradation of colony-stimulating factor 3 receptor (CSF3R) . The study by Meenhuis et al.  suggested that DUB-2A counteracts CSF3R ubiquitylation by the E3 ligase suppressor of cytokine signalling 3 at the endosome, inhibiting trafficking of CSF3R to the lysosome. However, its action could be alternatively explained if, similar to USP17, DUB-2A facilitates CSF3R CME, directing the receptor for recycling rather than degradation. However, DUB-2A localises to the plasma membrane , while USP17 localises to the endoplasmic reticulum, suggesting that these orthologues could function differently [64,65].
In addition to the DUBs involved in EGFR endocytosis, many other DUBs have also been shown to influence the endocytosis of multiple other receptors (Figure 2). Similar to USP9X, USP12 has an impact on Notch signalling by promoting Notch receptor degradation via the lysosome . USP12 silencing prevents trafficking of Notch to the lysosome, increasing its abundance at the cell surface and increasing Notch activity . USP12 can deubiquitylate Notch, but it is unclear how this contributes to Notch trafficking and further studies will be required to elucidate the mechanism utilised .
USP12 also regulates T-cell receptor (TCR) abundance at the cell surface. USP12 translocates from the nucleus to the cytosol upon TCR activation, and USP12 null cells have altered TCR signalling due to an attenuated level of TCR at the cell surface. USP12 deubiquitylates TCR adapter molecules, linker of activated T-cells (LAT) and T-cell receptor-associated transmembrane adapter 1 (Trat1), and prevents their lysosomal degradation. As the TCR rapidly cycles to and from the cell surface, and both LAT and Trat1 influence its cell surface stability, this indicates that USP12 stabilises the TCR at the cell surface by stabilising LAT/Trat1 .
Several DUBs have been implicated in the regulation of transforming growth factor β (TGF-β) receptor endocytosis (Figure 3). The TGF-β receptor has two subunits, TGF-β type 1 receptor (TβRI) and TGF-β type 2 receptor (TβRII), which complex upon TGF-β binding. TβRII internalises constitutively via CME and ligand binding recruits TβRI, and both subunits are then internalised to early endosomes where they signal by phosphorylating receptor-phosphorylated Smad proteins (R-Smads) recruited by Smad anchor for receptor activation . TGF-β receptor activation triggers the production of SMAD7, which can then bind to TβRI and recruit SMAD-specific E3 ubiquitin protein ligase 2 (Smurf2), an E3 ligase, which ubiquitinates TβRI and directs the receptor complex to lipid rafts where it undergoes caveolar-type endocytosis and traffics to the lysosome for degradation . USP4 interacts with TβRI and deubiquitylates it, counteracting its trafficking to the lysosome and stabilising it at the plasma membrane . USP4 normally localises to the nucleus, but TGF-β treatment triggers AKT phosphorylation of USP4 that causes its translocation to the cytosol, where a portion associates with the membrane and interacts with TβRI . USP15 also counteracts the ubiquitylation of TβRI by Smurf2, stabilises TβRI, and again enhances TGF-β signalling . USP15 binds to SMAD7 and deubiquitylates Smurf2, removing ubiquitin from Lys734, a residue required for Smurf2's ability to ubiquitylate TβRI and target it to the lysosome . USP15 also localises to the nucleus, but like USP4, it can shuttle to the cytosol allowing it to interact with TβRI . USP11 again interacts with SMAD7 and TβRI, as well as deubiquitylating TβRI to enhance TGF-β signalling . Interestingly, USP4, USP11, and USP15 are paralogous DUBs whose functions are at least partially redundant, something supported by their apparent regulation of TβRI ubiquitylation . UCH37 has also been shown to interact with SMAD7 and counteract the ubiquitination of TβRI by Smurf2. UCH37 again stabilises TβRI and boosts TGF-β signalling . Reduced levels of TβRI were also observed in the absence of UCH37, although this study suggested that UCH37 only affects the early phase of TGF-β signalling and only has an impact on TGF-β-dependent cell migration . However, it is unclear how UCH37 mediates these effects as it needs to interact with the proteasome to prompt activation , although it is possible that the proteasome has some role to play in these observations.
DUB regulation of TGF-β receptor endocytosis.
GPCRs are some of the most abundant receptors on cell surfaces and recent reviews have discussed the importance of ubiquitin in their endocytosis and trafficking . Several DUBs have also been implicated in the regulation of GPCR endocytosis (Figure 2). β-2 adrenergic receptors (β2ARs) are phosphorylated upon ligand engagement allowing them to bind to β-arrestin 2. The E3 ligase Nedd4 is then recruited by β-arrestin 2, and it ubiquitylates β2AR and triggers trafficking to the lysosome for degradation. Expression of both USP33 and USP20 impedes β2AR ubiquitylation and trafficking to the lysosome, and promotes its recycling to the plasma membrane . USP33 and USP20 were proposed to be functionally redundant, as both must be depleted to significantly decrease receptor recycling and resensitisation of β2AR . However, USP33 has subsequently been shown to interact with β2AR and β-arrestin 2 and to deubiquitylate β-arrestin 2 dissociating it from β2AR. This indicates that USP33 either blocks β-arrestin 2 recruitment and β2AR endocytosis, or causes its dissociation during endocytosis prompting recycling . It is also not clear when or where USP33 can influence β2AR endocytosis, as previous reports have localised USP33 to the Golgi apparatus . USP20 has alternatively been shown to be phosphorylated on serine 333 upon β2AR ligand engagement. This deactivates USP20 and promotes its dissociation from the activated β2AR complex, resulting in trafficking of the receptor for degradation .
The chemokine receptor CXCR4 (C-X-C chemokine receptor type 4), another GPCR, is ubiquitinated upon engagement with its ligand CXCL12 [stromal cell-derived factor 1 (SDF-1) also known as C-X-C motif chemokine 12], triggering trafficking to the lysosome for degradation. A range of DUBs were screened to determine which regulated CXCR4 down-regulation, and USP14 was shown to co-localise with CXCR4 at the cell surface and in endosomes upon CXCL12 engagement and to deubiquitylate this receptor . However, determining the exact role of USP14 has been complicated by the observation that overexpression and knockdown of USP14 both block CXCR4 down-regulation and CXCL12-driven cell migration. This suggests that both addition and removal of ubiquitin are required to allow CXCR4 trafficking to the lysosome . The role of USP14 is complicated further by previous work (Ubiquitinated proteins activate the proteasome by binding to Usp14/Ubp6, which causes 20S gate opening’. This is in Molecular Cell in 2009, volume 36, Pages 794–804 and is authored by Andreas Peth, Henrike C. Besche and Alfred L. Goldberg.) which indicates it is associated with the proteasome, suggesting that it may also play a role in this process.
As mentioned above, USP17 is required to facilitate CME, not only for EGFR, but potentially a range of different receptors. However, it is not the only DUB which potentially favours specific endocytosis routes. In yeast, UBP2 and UBP7 are proposed to facilitate CME as their deletion increases the lifetime of early endocytic coat proteins at the plasma membrane . However, the deletion of both does not block CME, only slows its progress, suggesting that they are not vital for its conclusion . UBP2 and UBP7 have subsequently been shown to deubiquitylate Ede1 (EH domain-containing and endocytosis protein 1), a homologue of mammalian Eps15, which is deubiquitylated by USP9X . Interestingly, like USP9X, UBP2/7 also have post-internalisation roles, particularly in endosomal sorting (discussed later), and it is possible that this also contributes to their impact on CME.
DUBs and ion channel endocytosis
Transmembrane receptors are not the only cargoes whose function and availability at the plasma membrane are regulated by endocytosis; many ion channels have also been shown to undergo endocytosis to regulate their function (Figure 4).
DUB regulation of ion channel endocytosis.
The epithelial sodium channel (ENaC), vital for sodium homeostasis, is ubiquitylated targeting it to the lysosome. ENaC ubiquitylation is counteracted by USP2b, which reverses its down-regulation and stimulates Na+ reabsorption . USP2b was initially thought to interact directly with ENaC, but subsequent studies suggest that it interacts with Nedd4-2, the E3 responsible for its ubiquitylation . However, the physiological relevance of USP2b-dependent regulation of ENaC remains to be clarified, as USP2 null mice do not show any evidence of defects in the regulation of ENaC or sodium transport in the kidney .
KCNQ1 potassium channels are also ubiquitylated by Nedd4-2 and targeted for down-regulation by lysosomal degradation, and again these channels are deubiquitylated and stabilised by USP2b . Voltage-gated calcium channels (Cav) are ubiquitylated by the related E3 ligase Nedd4-1, and USP2b associates with the α2δ-1 accessory subunit that is required for Cav1.2 down-regulation. However, while USP2b does deubiquitylate both Cav1.2 and α2δ-1, in contrast with the other channels it favours their degradation and down-regulation .
Cystic fibrosis transmembrane conductance regulator (CFTR) is a chloride channel that helps maintain the volume of liquid on the surface of the airways. CFTR abundance at the apical plasma membrane of bronchial epithelial cells is regulated by endocytosis and recycling to the membrane . USP10 interacts with and deubiquitylates CFTR, rescuing it from lysosomal degradation and favouring CFTR recycling to the apical plasma membrane . USP10 also increases ENaC levels at the cell surface, although, unlike CFTR, it does not directly deubiquitylate ENaC . Instead, USP10 was found to have an impact on ENaC endocytosis by deubiquitylating sorting nexin 3 (SNX3) to rescue it from proteasomal degradation .
Finally, similar to USP10, UCH-L3 depletion increases ENaC ubiquitination and reduces its levels at the apical plasma membrane, as well as decreasing ENaC-mediated Na+ currents. This indicates that UCH-L3 can rescue ENaC from lysosomal degradation . However, it is not clear how it does this as recent studies suggest that UCH-L3 is not capable of hydrolysing the ubiquitin chain types attached to ENaC and further investigation is required to elucidate the exact mechanism responsible for these observations .
DUBs and central nervous system endocytosis
Endocytosis within the central nervous system (CNS) is of particular importance due to the need for the generation of new SVs after neurotransmitter release is triggered and existing SVs fuse with the plasma membrane. Indeed, cells in the brain are constantly cycling their plasma membranes to facilitate their function, and endocytosis of CNS receptors, such as α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPARs), can influence CNS function [93–95]. However, endocytosis within nerve terminals appears to operate differently from other cells and can be influenced by the level of stimulation in an individual neurone. In particular, low-level stimulation triggers ultrafast endocytosis that allows rapid SV retrieval, while stronger stimulation favours slower CME, and high-intensity stimuli which saturate these modes of SV retrieval can trigger activity-dependent bulk endocytosis that retrieves large regions of the plasma membrane. It is thought the internalised membranes form into endosomes from which SVs can be generated, although there are still controversies about the form and existence of these endosomes [96,97]. As a result of these differences and the difficulties of generating appropriate models to study CNS endocytosis, our knowledge in this area has lagged behind that for other tissues. However, a better understanding of these processes is now being elucidated and this has included studies examining the impact of DUBs on the endocytosis of CNS receptors.
AMPAR accumulation at synapses is thought to contribute to higher brain functions such as learning and memory . AMPARs are multisubunit ion channels [99,100] and glutamate binding is thought to trigger AMPAR ubiquitylation, endocytosis and trafficking to the lysosome , although a recent study indicates that it can also be degraded by the proteasome . USP46 was originally shown to regulate AMPAR trafficking in Caenorhabditis elegans  where it deubiquitylates the glutamate receptor-1 (GLR-1) subunit, preventing its degradation [94,95] and stabilising it at the synapse . USP46 is also enriched in rat synapses where it co-localises with the AMPAR subunit GluA1 and deubiquitylates it, impeding its degradation and increasing AMPAR-mediated synaptic transmission . USP46 knockout mice also have impaired long-term memory, reinforcing the importance of USP46 in higher brain function , and mice with a deletion in USP46 that removes a lysine codon also show changes in the tail suspension test, which is widely used to assess depression-like behaviour. These studies indicate that USP46 also regulates the γ-aminobutyric acidergic (GABAergic) system, although how this is accomplished is still to be elucidated [106,107].
Glutamate transporters also regulate glutamate signalling by removing extracellular glutamate, with the main glutamate transporter, GLT-1, clearing ∼90% of extracellular glutamate in the cerebral cortex . GLT-1 constitutively internalises via CME and recycles to the plasma membrane. UCH-L1 is highly expressed in neurones and an UCH-L1 inhibitor blocks GLT-1 recycling, indicating that it plays a role in the recycling of this receptor . UCH-L1 mutations are associated with Parkinson's disease, suggesting a role in the regulation of protein folding and turnover in the CNS. However, it is unclear how UCH-L1 affect GLT-1, and as the impact of the UCH-L1 inhibitor is the only evidence to date supporting its role, questions still remain as to whether this inhibitor may also affect other enzymes.
AxJ mice display an ataxic phenotype due to a marked reduction in the amount of full-length USP14 expressed in the brain. AxJ mice have increased cell surface expression of GABAA receptors (GABAARs) as well as increased GABAergic current amplitudes. This indicates that USP14 is a negative regulator of GABAAR surface expression and USP14 interacts and co-localises with GABAAR at the synapse. GABAAR α1 is ubiquitylated and expression of an USP14 mutant increases its ubiquitylation. However, USP14 overexpression has little effect, suggesting that it does not directly deubiquitylate GABAAR .
USP14 also interacts and co-localises with GABABR, which is again ubiquitylated prompting its internalisation and degradation. USP14 depletion reduces the ubiquitylation and increases the half-life of GABABR, while USP14 expression accelerates GABABR degradation, suggesting that USP14 favours ubiquitylation and degradation of this receptor. However, the impact of USP14 does not require its DUB activity, suggesting an alternative mode of action . As mentioned before, USP14 is associated with the proteasome, but it is unclear how this contributes to USP14's regulation of GABAAR and GABABR, and further work will be required to try to elucidate exactly what is going on here.
DUBs and endosomal sorting
Ubiquitylation of many receptors and ion channels results in their endocytosis and trafficking to the lysosome via early endosomes and MVBs. However, until they are incorporated into the vesicles that invaginate into the lumen of MVBs, they are not irreversibly committed to degradation within the lysosome and DUBs can still act to regulate this process. Trafficking of cargo into these luminal vesicles is facilitated by the machinery of the endosomal sorting complexes required for transportation (ESCRT) system . The first point of contact for an internalised ubiquitylated receptor is with ESCRT-0 components, which comprises hepatocyte growth factor-regulated tyrosine kinase substrate (HRS) and signal transducing adaptor molecule (STAM), both of which contain ubiquitin-interacting motifs . This is then followed by transfer of the ubiquitylated receptors downstream to multimeric ESCRT-I and ESCRT-II complexes . ESCRT-III components and the AAA–ATPase VPS4 (vacuolar protein sorting-associated protein 4) then conclude the process by generating internal vesicles . The ESCRT system is central to endosomal trafficking and is conserved from yeast to mammalian cells.
In yeast, several DUBs have been implicated in endosomal sorting. Doa4 was originally proposed to act late in the MVB pathway and deubiquitinate cargo proteins, or components of the ESCRT machinery, to facilitate proper sorting to the lysosome . Recently, Doa4 has been shown to interact with the ESCRT-III component Snf7 (vacuolar-sorting protein 7) and preferentially cleave K63-linked ubiquitin chains involved in membrane protein trafficking . UBP2 and UBP7 interact with Hse1, the yeast orthologue of STAM, but have opposite effects on sorting. UBP7 binds directly to Hse1 and its loss increases the efficiency of sorting, suggesting that it counteracts this process . UBP2 interacts with the E3 ligase Rsp5, which binds to Hse1, and this complex appears to be necessary for sorting . This combined with their previously discussed impact on CME indicates that UBP2 and UBP7 play an integral role in receptor endocytosis and sorting in yeast, although how, and if, these two roles are linked, is currently unclear.
In mammals, AMSH (associated molecule with SH3 domain of STAM) and USP8 (UBPy) both localise to endosomes/MVBs and both interact with STAM, indicating a role in endosomal sorting [115,116]. AMSH is a JAMM family member that is highly selective for K63-linked polyubiquitin chains, while USP8 has no apparent chain selectivity [117,118]. USP8 binding stabilises the HRS/STAM complex, indicating that it is vital for proper sorting , and AMSH activity has recently been shown to be enhanced by STAM binding confirming its role in this complex [120,121]. However, AMSH and USP8 may not simply regulate ESCRT-0 as both have also been shown to interact with ESCRT-III via N-terminal MIT domains [17,122], and AMSH can bind to clathrin, an interaction which localises it to the early endosomes [26,123]. Indeed, how AMSH and USP8 contribute to the regulation of endosomal sorting and degradation of receptors has proved to be controversial with several different theories being proposed.
AMSH was originally thought to deubiquitinate EGFR and oppose its sorting into the luminal vesicles of the MVBs due to the observation that AMSH depletion enhanced EGFR lysosomal degradation . However, subsequent studies have challenged this, demonstrating that AMSH must be localised to the endosome to allow EGFR degradation . In addition, the differences in EGFR and Erb-B receptor tyrosine kinase 2 (ErbB2) recycling correlate with the level of deubiquitylation by AMSH, indicating that AMSH deubiquitylation opposes recycling. In addition, stimulating EGFR with TGF-α, a ligand which favours EGFR recycling, caused efficient K63 polyubiquitylation of EGFR, but AMSH-mediated deubiquitination was significantly reduced. As a result, it was concluded that recycling is associated with significantly reduced deubiquitylation by AMSH, indicating that AMSH favours trafficking to the lysosome . This has been further supported by studies examining many GPCRs, including CXCR4, protease-activated receptor 2 (PAR2) and the δ-opioid receptor (DOR). Expression of a dominant-negative AMSH mutant, or AMSH knockdown by siRNA, increased steady-state levels of CXCR4 and blocked its degradation upon CXCL12 stimulation . Similar experiments were carried out examining PAR2 and DOR, and again AMSH was observed to favour their degradation [127,128]. As a result, currently the evidence would support a role for AMSH in facilitating receptor turnover by favouring trafficking to the luminal vesicles of the MVB and degradation in the lysosome (Figure 5).
DUB regulation of endosomal sorting.
The role of USP8 (UBPY in mice) is also controversial with an initial study reporting that USP8 depletion increased EGFR degradation, suggesting that USP8 opposed EGFR trafficking to the lysosome . However, subsequent studies have indicated that cells lacking USP8 accumulate ubiquitylated proteins on endosomes, as well as exhibiting an increase in the number and size of MVBs. They also note that EGFR and hepatocyte growth factor receptor (MET or HGFR) degradation is inhibited, suggesting that USP8 is essential for receptor down-regulation . This is supported by studies indicating that USP8 deubiquitylation promotes EGFR degradation  and studies examining USP8 null mice. These mice are embryonic lethal, but conditional USP8 knockouts show that USP8 null cells have enlarged endosomes, more MVBs, and ubiquitylated substrates accumulate on these enlarged endosomes. USP8 depletion also leads to the destabilisation of the HRS/STAM complex providing a potential explanation for these observations. However, surprisingly, USP8 null cells were also found to exhibit reduced levels of EGFR, MET, and ErbB3, indicating that USP8 is actually required for their stability . Another study has again indicated that USP8 depletion accelerates EGFR degradation, suggesting that USP8 opposes EGFR lysosomal trafficking . However, this study also observed that a dominant-negative USP8 mutant caused the accumulation of a hyperubiquitylated EGFR in early endosomes, which would alternatively suggest USP8 is required for EGFR degradation via the MVBs and lysosome .
The role of USP8 with regard to EGFR stability may not be clear cut; however, the general consensus for other receptors indicates that USP8 is a positive regulator of receptor lysosomal degradation (Figure 5). In particular, USP8 depletion prevents the lysosomal trafficking and degradation of PAR2 , CXCR4 , vascular endothelial growth factor receptor 2 , LDLR , and the Ca2+-activated K+ channel (KCa3.1) . Indeed, ENaC is the only example where USP8 opposes degradation, with USP8 deubiquitylation of ENaC preventing its degradation, favouring recycling to the plasma membrane, and increasing sodium currents  (Figure 5).
In addition to its role at the MVB, many other explanations have been proposed for the impact of USP8 on the endocytosis of particular plasma membrane proteins. The seven transmembrane protein Smoothened (Smo) is a critical component of the Hedgehog (Hh) signalling pathway in Drosophila and its accumulation at the cell surface is required for Hh signalling. Smo is multi-monoubiquitylated, and this mediates its endocytosis and trafficking to the lysosome for degradation. USP8 has been shown to deubiquitylate Smo causing its accumulation at the cell surface increasing Hh signalling [138,139]. USP8 is also highly expressed in the brain , and it deubiquitylates AMPARs increasing their cell surface expression. In particular, USP8 was found to oppose the action of the E3 ligase Nedd4-1, which ubiquitylates AMPARs prompting their lysosomal trafficking and degradation [95,141]. USP8 also deubiquitylates and stabilises the E3 ligase RNF41 (RING finger protein; Nrdp1), which ubiquitylates ErbB3 and ErbB4 and regulates their levels . Indeed, the ErbB3 ligand neuregulin-1 was found to trigger USP8 phosphorylation, stabilise RNF41 in an USP8-dependant manner, and promote ErbB3 ubiquitylation and down-regulation . However, RNF41 also blocks the cathepsin-L cleavage of multiple JAK2-dependent cytokine receptors [leptin receptor (LR), leukaemia inhibitory factor receptor (LIFRα), and IL-6 receptor (IL-6α)], which triggers their trafficking to the lysosome for degradation . USP8 depletion reduces the lysosomal degradation of LR and LIFRα and enhances their recycling and subsequent shedding, which resembles the impact of RNF41 overexpression. This, in contrast with the findings for ErbB3/4, indicates that USP8 actually antagonises the action of RNF41, rather than stabilising and enhancing its function, and is a positive regulator of cytokine receptor trafficking to the lysosome for degradation. However, as it is not completely clear how RNF41 and USP8 regulate the cathepsin-L cleavage, further work is required to clarify what exactly is going on . It is also unclear how endosome localised USP8 is able to influence events at the cell surface and it is possible that USP8 may mediate its role from the endosome/MVB.
The relationship between USP8 and the lysosome has also recently been further complicated by the observation that USP8 plays a role in the maintenance of lysosomes. Lysosomal enzymes, such as cathepsins, are trafficked from the trans-Golgi network (TGN) to the endosomes, and then the lysosomes, as cargo bound to the cation-independent mannose 6-phosphate receptor (ci-M6PR) . ci-M6PR is then recycled back to the TGN via the retromer pathway before picking up more cargo. USP8 depletion redistributes the bulk of ci-M6PR from the TGN to the endosomes and disrupts cathepsin D trafficking, causing it to remain in its unprocessed form and to be secreted from the cell. This indicates that USP8 is required for retromer-mediated ci-M6PR recycling to the TGN, a process previously shown to involve HRS through its interaction with the retromer components VPS35 and SNX1 . It also indicates that USP8 depletion blocks trafficking of cathepsins to the lysosome, as ci-M6PR fails to recycle to the TGN to pick them up (Figure 5) . This is very interesting as it raises the possibility that defective delivery of enzymes to the lysosome also contributes to the failure of receptor turnover in cells lacking USP8  and could also potentially account for the previously observed impact of USP8 on the cathepsin-L cleavage of cytokine receptors .
USP7 has also recently been linked to retromer-mediated trafficking of ci-M6PR (Figure 5). The MAGE-L2-TRIM27 E3 ligase promotes endosomal protein recycling by K63-linked polyubiquitylation of WASH, which is essential for this pathway. USP7 interacts with MAGE-L2-TRIM27 and prevents TRIM27 auto-ubiquitylation and degradation. Indeed, USP7 depletion mimicked MAGE-L2, TRIM27, or WASH knockdown, impairing ci-M6PR recycling and resulting in its endosomal accumulation . Previously, USP7 has been linked to multiple functions and shown to localise to the nucleus , but this study indicated that it could also co-localise with TRIM27 in endosomes .
The current consensus with regard to endosomal sorting would suggest that in the majority of cases, both AMSH and USP8 favour the incorporation of receptors into the vesicles of the MVBs and ultimately lysosomal degradation. However, the variation in observations by different groups, especially with regard to EGFR, could well result from differences in the cell models and conditions used in each case. In addition, the fact that AMSH and USP8 also potentially act at multiple points in this sorting cascade could contribute to these contradictory observations, as well as USP8's potential role in the generation of lysosomes which may muddy the waters further. As a result, although a consensus view has been established, there are many questions that still remain in regard to this system.
DUBs endocytosis, endosomal sorting, and disease
Many of the DUBs mentioned in this review are implicated in the progression of multiple diseases. However, these DUBs have multiple functions, many of which are unrelated to receptor endocytosis and trafficking. As a result, here we will focus on those diseases where their role is, or is likely to be, related to receptor endocytosis and trafficking.
Mutations of AMSH, which is involved in endosomal sorting, have been found in patients with microcephaly–capillary malformation syndrome. Cells from these patients display reduced AMSH expression, accumulation of ubiquitin-conjugated aggregates, increased apoptosis, and aberrant intracellular signalling. The exact mechanism underlying these effects is not completely clear; however, the accumulation of ubiquitin conjugates indicates that it is likely to be associated with a failure to properly sort proteins for lysosomal degradation .
A significant number of pituitary adenomas in Cushing's disease (CD) sufferers have recently been shown to exhibit USP8 mutations that impair USP8 binding to 14-3-3 proteins and enhance its deubiquitylating activity . CD results from pituitary adenomas that secrete adrenocorticotropic hormone (ACTH). How these USP8 mutations elevate ACTH secretion is not completely clear, but it could well involve aberrant endosomal sorting and protein secretion. EGFR activation results in the production of proopiomelanocortin, a precursor of ACTH , and many adenomas with USP8 mutations have higher EGFR levels, suggesting that regulation of EGFR could drive ACTH overproduction [132,149]. However, this is not always the case and other mechanisms may be responsible .
Heterozygous USP7 mutations have been observed in several patients displaying neurodevelopment disorders. These were highlighted in the study examining the role of USP7 in retromer-mediated trafficking, and it was noted that there was overlap between the symptoms observed and those of Schaaf-Yang syndrome, a disease caused by MAGE-L2 loss of function. This indicates that USP7 WASH regulation is important in the CNS and that the neurodevelopment disorders observed result from defects in retromer trafficking .
As described above, USP46 regulates AMPAR endocytosis , and mice with an USP46 mutation show signs of despair, apparently due to a defect in GABAergic signalling . Although no USP46 mutations have been identified, studies indicate an association between USP46 and major depressive disorder in the Japanese population . In addition, one family which suffers from early onset essential tremor has a variant of USP46 that is associated with this disease, indicating a potential role for USP46 .
Finally, amplifications of the USP15 gene are observed in glioblastoma patients and are associated with poor prognosis. This is proposed to result from its effects on TGF-β signalling and USP15 depletion in orthotopic mouse models of glioblastoma has been shown to decrease their oncogenic potential and their TGF-β activity .
This is a fairly brief list of DUBs whose role in endocytosis contributes to disease; however, as our understanding of how DUBs contribute to the endocytosis of more receptors starts to crystallise, and efforts to sequence large cohorts of patients suffering from a range of diseases gather pace, it is likely that many additional examples will be added to this list as we go forward.
In recent years, our knowledge with regard to the role of ubiquitin has expanded hugely, along with our knowledge of the machinery required for ubiquitylation and deubiquitylation. As part of this, DUBs have come to the fore as a fascinating group of proteases which have many unusual features. Initially, unlike most other protease families, all DUBs were thought to cleave the covalent bond between ubiquitin and the ε-amino group of lysine residues on target proteins, or other ubiquitin molecules. Indeed, it was thought that they all could cleave the same substrates and any specificity they displayed was the result of their ability to bind to a subset of substrates via the other domains which they incorporated. However, it has now become apparent that DUBs display specificity for the different polyubiquitin chain types (linear, K6, K11, K27, K29, K33, K48, and K63), or for ubiquitin-like molecules, and in addition to their other domains, this contributes to their substrate selection . In addition, again unlike many other protease families, endogenous inhibitors of DUBs have as yet to be identified, suggesting that their activity is regulated by alterations in transcription, translation, or their post-translational modifications. The apparent lack of specificity, as well as the absence of endogenous inhibitors, has hindered studies of individual DUBs as specific inhibitors have proved difficult to identify. As a result, studies have had to rely on overexpression, knockdown, or knockout experiments to elucidate the roles these various DUBs perform, something which does not always suit the study of dynamic processes such as endocytosis.
Even with these difficulties, over the last decade, our understanding of the potentials roles DUBs play in the regulation and control of endocytosis, endosomal sorting, and receptor degradation has expanded hugely. However, there are still many questions that need to be resolved with regard to exactly how the DUBs implicated actually contribute to the regulation of these processes. In addition, there are still many receptors whose endocytosis has not yet been fully investigated. Therefore, it is likely that the influence of DUBs in this field is only set to expand over the years to come.
On first glance, there do not appear to be many common threads running through the DUB regulation of endocytosis. Indeed, a plethora of different DUBs have been implicated in the regulation of the endocytosis, recycling, and degradation of a wide range of receptors and ion channels, and even where the same DUB is involved, it does not always appear to utilise the same mode of action [46,47,58,59]. This could be due to the fact that most endocytosis studies concentrate on the likes of EGFR and TfR; thus, our understanding of what is happening with other receptors is not as well developed. It could also result from the range of cell and animal models being utilised, which potentially differ fundamentally in their regulation of endocytosis. The mode of action may also be confused due to the propensity of DUBs to deubiquitylate multiple substrates, and thus altering one substrate could affect the regulation of others. There may also be some confounding factors, such as the tendency of DUBs that associate with the proteasome (UCH37 and USP14) [75,82] to affect the degradation of multiple substrates and multiple processes indirectly.
However, despite these issues, there are several common threads which are of interest. USP2 and USP9X regulate the internalisation and recycling of a range of receptors [46,47,52,58,59,62] and ion channels [85,87,88], suggesting that they potentially play general roles in this process. It is also likely, even though the roles that have been proposed do not always match, these DUBs play a common role in all cases [46,47,58,59]. DUBs also appear to be capable of facilitating specific routes of endocytosis (e.g. USP17 and JosD1) [54,84]. In particular, USP17 facilitates CME of EGFR and TfR, as well as being induced in response to the activation of multiple receptors that potentially undergo CME . USP46 also appears to be of importance in the CNS, with multiple CNS receptor systems potentially being regulated by this DUB [103–105]. In addition, to facilitate their roles in endocytosis, several DUBs translocate from the nucleus to the cytosol to allow them to interact with the receptor they regulate [68–70,73]. This is a particular feature of the paralogous DUBs that regulate TβRI ubiquitylation and trafficking (USP4, USP11, and USP15) [69,70,73], although USP12 also moves to the cytosol upon TCR stimulation allowing it to regulate TCR abundance at the plasma membrane . It is also clear that both AMSH and USP8 are pivotal for endosomal sorting, although there is still some debate as to the exact details of their action. Future studies will likely strengthen these threads and potentially reveal other DUBs, which play significant roles in these processes. In addition, as our understanding of endocytosis in the CNS progresses, it is likely that other DUBs will be implicated in the regulation of these processes at the synapse.
Finally, although we have only discussed a few examples of DUBs whose role in endocytosis are linked to disease, as our knowledge of these pathways expands it is likely that other disease links will be established. In addition, as the endocytosis of receptors and channels facilitates their proper function, it is likely that DUBs, which positively or negatively influence the endocytosis of these cargoes, will be identified as novel and exciting targets for pharmaceutical intervention to modulate the function of disease-associated receptors and channels. In addition, although DUBs have proved to date to be elusive pharmacological targets, recent studies suggest that specific inhibitors can be synthesised and this may intensify the interest in these proteases and mark them as a promising therapeutic avenue in the future . As a result, our interest in the DUBs that play a role in the regulation, or facilitation, of endocytosis and endosomal trafficking only seems set to increase.
α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor
α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors
associated molecule with SH3 domain of STAM
voltage-gated calcium channel
cystic fibrosis transmembrane conductance regulator
cation-independent mannose 6-phosphate receptor
central nervous system
colony-stimulating factor 3 receptor
stromal cell-derived factor 1 (SDF-1) also known as C-X-C motif chemokine 12
C-X-C chemokine receptor type 4
epidermal growth factor receptor
epithelial sodium channel
epidermal growth factor receptor pathway substrate 15
Erb-B receptor tyrosine kinase 2
endosomal sorting complexes required for transportation
glutamate receptor 1 or GluA1
glutamate transporter 1
hepatocyte growth factor-regulated tyrosine kinase substrate
ligase inducible degrader of the LDLR
linker of activated T-cells
low-density lipoprotein receptor
leukaemia inhibitory factor receptor
monocyte chemotactic protein-induced protein
hepatocyte growth factor receptor (HGFR) or MET
Machado-Joseph disease protein domain protease
ovarian tumour protease
RING finger protein
SMAD-specific E3 ubiquitin protein ligase 2
vacuolar-sorting protein 7
signal transducing adaptor molecule
TGF-β type 1 receptor
TGF-β type 2 receptor
transforming growth factor β
T-cell receptor-associated transmembrane adapter 1
ubiquitin C-terminal hydrolase
vacuolar protein sorting-associated protein
β-2 adrenergic receptors
Our research upon deubiquitinating enzymes has been funded by Action Cancer (Project grant), Biotechnology and Biological Sciences Research Council [BB/F013647/1] and Queen's University Belfast.
We apologise to those researchers whose work we could not cite due to space constraints.
The Authors declare that there are no competing interests associated with the manuscript.