Many neurodegenerative diseases are associated with aggregate-prone neurotoxic proteins (aggresomes or inclusion bodies) that perturb cellular homeostasis and neuronal function (Popovic et al., 2014). These diseases are also categorically classed as proteinopathies, and comprise aggregation-prone components such as β-amyloid (Aβ) and tau in Alzheimer’s disease (AD), α-synuclein in Parkinson’s disease (PD), and polyglutamine (polyQ) repeat diseases.

Alzheimer’s disease is the most common form of progressive neurodegeneration and represents the main cause of senile dementia affecting about 10% of the population over the age of 65 and about 50% of the population over 85. AD can be divided into two forms: familial AD and sporadic AD. Familial AD accounts for 5%∼10% of AD patients, where inherited autosomal dominant mutations in three genes have been considered as the primary causes of familial AD: these include mutations in the genes encoding β-amyloid precursor protein (APP) and presenilins (PS1 and PS2; Goate et al., 1991; Levy-Lahad et al., 1995; Sherrington et al., 1995). In addition, sporadic AD is associated with polymorphisms in the Apolipoprotein E (APOE) gene, where the epsilon four allele is a strong genetic risk factor for late-onset AD (Strittmatter et al., 1993; Bu, 2009; Fu et al., 2016). Accumulation and aggregation of neurotoxic proteins, such as β-amyloid (Aβ), hyperphosphorylated tau, ubiquitinated proteins, and other unfolded proteins in vulnerable brain regions such as the hippocampus and cortex in AD brain is central to disease pathogenesis (Selkoe, 2001). Two major pathological hallmarks in AD are extracellular amyloid plaques and intraneuronal neurofibrillary tangle (NFT; Zhang et al., 2011). NFTs are comprised of hyperphosphorylated microtubule-associated protein tau (Song et al., 2015; Yamada et al., 2015; Zempel and Mandelkow, 2015). Amyloid plaques are primarily comprised of Aβ which is derived from sequential cleavages of APP by β- and γ-secretases (Zhang et al., 2011; Audrain et al., 2016). Cumulative evidence demonstrates that Aβ and tau are neurotoxic and can trigger a cascade of neurodegenerative processes ending in neuronal death, suggesting that overproduction/accumulation of Aβ and tau in vulnerable brain regions is the primary influence driving AD pathogenesis (Hardy and Selkoe, 2002; Gautam et al., 2015; Golovyashkina et al., 2015; Popugaeva et al., 2015).

Parkinson’s disease is the second most common neurodegenerative disease, clinically characterized by motor abnormalities, including resting tremor, rigidity, hypokinesia, and postural instability (Dauer and Przedborski, 2003; More and Choi, 2015). The loss of dopaminergic neurons in the substantial nigra subregion of midbrain is the major cause of motor deficits (Chen et al., 2015). Major pathological features of PD include intraneuronal accumulation of Lewy bodies and dystrophic neurites (Lewy neurites), both of which comprise aggregated proteins, such as α-synuclein, parkin, and ubiquitinated proteins (Hughes et al., 1992; Braak et al., 2003; Chiasserini et al., 2015; Xiang et al., 2015). Sporadic PD accounts for more than 90% of PD cases (Lang and Lozano, 1998), and only less than 10% of familial PD cases are caused by monogenic mutations (Benitez et al., 2016). Mutations in the gene encoding α-synuclein (Polymeropoulos et al., 1997; Helferich et al., 2015; Majbour et al., 2016; Schonherr et al., 2016) represent autosomal dominant familial PD, and mutations in genes encoding parkin (Kitada et al., 1998; Cha et al., 2015), UCHL1 (Leroy et al., 1998; Maraganore et al., 1999), DJ-1 (Bonifati et al., 2003), PINK1 (Valente et al., 2004), and LRRK2 (Zimprich et al., 2004; Saha et al., 2015) cause autosomal recessive familial PD.

Huntington’s disease (HD) is an autosomal dominant progressive neurodegenerative disorder, clinically characterized by cognitive impairment, chorea and dystonia, bradykinesia, uncoordinated movement, mood disorder and other psychiatric symptoms, and behavioral difficulties (Huntington Study Group, 1996). As the most common of the polyQ disorders, HD is caused by expansion of a CAG repeat coding for polyQ in the N-terminus of the huntingtin protein (MacDonald et al., 1993; Evers et al., 2015; Nekrasov et al., 2016). There is a remarkable repeat length threshold effect, where polyQ chain expansion of 36 repeats triggers HD onset, whereas repeats of less than 35 are non-pathological (Chong et al., 1997). The polyQ expansion in the N-terminus of huntingtin protein confers a gain-of-function phenotype, and is associated with significant neurotoxicity (Tallaksen-Greene et al., 2003).

In addition to huntingtin, expansion of polyQ-encoding CAG repeats in several genes have also been found in other polyQ repeat diseases, such as Ataxin-1 in spinocerebellar ataxia type 1 (SCA1), Ataxin-2 in SCA2, Atantin-3 in SCA3, α1_A-voltage-dependent calcium channel (CACNA1A) in SCA6, Ataxin-7 in SCA7, Atrophin-1 in dentatorubral-pallidoluysian atrophy (DRPLA), and androgen receptor in SBMA (Zoghbi and Orr, 2000; Del Cano-Espinel et al., 2015). All polyQ repeat disorders are dominant neurodegenerative diseases and are associated with pathological intracellular inclusion bodies in the affected brain regions (Ross et al., 1997; Zoghbi and Orr, 2000).

Ubiquitin is an evolutionarily conserved 76-amino acid moiety covalently tandemly linked to target protein components for degradation by the ubiquitin-proteasome system (UPS), and is required for degradation of about 80% intracellular proteins in eukaryotes (Pickart, 2001). The presence of ubiquitin in intracellular inclusions has been found in various neurodegenerative diseases, for example, amyloid plaques and NFT in AD, Lewy bodies in PD, and intranuclear inclusions in polyQ disorders (Todi and Paulson, 2011).

Protein ubiquitination occurs through the coordinated activity of several enzymes, including an ubiquitin activating enzyme (E1 ligase), conjugating enzyme (E2 ligase), and E3 ligase. Initially, a single ubiquitin moiety is attached to an active-site Cysteine residue within the E1 ligase through a thioester bond in an ATP dependent manner. The activated ubiquitin is then transferred to an ubiquitin conjugating enzyme-E2 ligase. Finally, the E2 ligase will cooperatively transfer the ubiquitin chain with a specified E3 ligase to a particular substrate, whereby the 26S proteasome will target the polyubiquitinated protein for degradation (Figure 1).

Ubiquitin E3 ligases determine the specificity of ubiquitination process. E3 ligases are grouped into three classes according to their unique domains, including Really Interesting New Gene (RING) or U-box domain-containing E3 ligases, Homologous to E6AP C-terminus (HECT) domain-containing E3 ligases, and RING-between-RING (RBR) domain-containing E3 ligases (Atkin and Paulson, 2014; Morreale and Walden, 2016). RING E3 ligases catalyze the transfer of ubiquitin directly from E2 ligases to the substrates by binding both of them. Whereas HECT E3 ligases comprise a conserved Cysteine residue as an ubiquitin acceptor, where the activated ubiquitin is initially transferred from the E2 ligases to HECT E3 ligases prior to E3-mediated substrate conjugation (Ravid and Hochstrasser, 2008). Unlike RING E3 ligases, RBR E3 ligases contain two RING domains (RING1 and RING2) separated by an in-between-RING (IBR) domain. Functionally, RBR E3 ligases are similar with HECT E3 ligases, both catalyze ubiquitin first transfer from E2 ligases to a catalytic Cysteine residue on E3 ligases then from the E3 ligases to the substrates (Spratt et al., 2014) (Figure 1).

The fate of an ubiquitin-linked protein is dependent on Lysine chain-specific polyubiquitination (Grabbe et al., 2011; Kulathu and Komander, 2012; Rieser et al., 2013). For example, Lys48-linked polyubiquitin chains usually target proteins for degradation (Husnjak et al., 2008), whereas Lys63-linked polyubiquitination is involved in regulation of NFκB signaling (Laplantine et al., 2009), Lys27-linked polyubiquitin chains are important for mitophagy (Geisler et al., 2010), and Lys11-linked polyubiquitination is implicated in cell cycle regulation (Bremm and Komander, 2011). Therefore, ubiquitination is not only important for protein degradation, but is also implicated in non-proteolytic functions, such as endocytosis, protein-protein interaction, intracellular trafficking, inflammatory signaling, autophagy, and DNA repair (Grabbe et al., 2011; Kulathu and Komander, 2012; Rieser et al., 2013).

The UPS is involved in protein quality control and removal of misfolded and aggregated proteins, and dysfunction of the UPS is implicated in the pathogenesis of neurodegenerative diseases (Popovic et al., 2014).

The process of ubiquitination is reversible and the reverse process is called deubiquitination which is accomplished by deubiquitinating enzymes (DUBs; Rubinsztein, 2006; Reyes-Turcu et al., 2009; Popovic et al., 2014). Increasing evidence indicates the importance of DUBs in regulating UPS function, including proofreading ubiquitin-protein conjugates, removing ubiquitin from conjugated proteins as key examples (Amerik and Hochstrasser, 2004) (Figure 1). Pivotal roles for DUBs have been demonstrated in maintaining neuronal functions and regulating the pathogenesis of various neurodegenerative diseases (Rubinsztein, 2006; Todi and Paulson, 2011; Popovic et al., 2014).

More than 100 DUBs have been identified (Popovic et al., 2014), and can be subdivided into five subfamilies. DUBs can be further categorized into two classes, the Cysteine proteases and the zinc-dependent metalloproteases (Todi and Paulson, 2011). Cysteine protease-associated DUBs include the ubiquitin carboxyl-terminal hydrolases (UCHs), ubiquitin specific proteases (USPs), Machado Joseph Diseases (MJD) proteases, and ovarian tumor (OTU) protease. Metalloprotease DUBs comprise the JAB1/MPN/Mov34 metalloenzyme (JAMM) subfamily.

UCHL1, a member of the UCH subfamily, is one of the most abundant proteins in the brain. UCHL1 is known to interact with and stabilize monomeric ubiquitin, and facilitates the E3 ligase activity in its dimeric form (Doran et al., 1983; Wilkinson et al., 1989). An Aplysia UCHL1 ortholog, ubiquitin carboxyl-terminal hydrolase (AP-Uch) has been found to be important in maintaining synaptic function (Hegde et al., 1997). Inhibiting the expression or function of AP-Uch blocks induction of long-term facilitation, which is thought to be the molecular basis for learning and memory (Hegde et al., 1997). Mammalian UCHL1 is required for normal synaptic structure and function in hippocampal neurons (Cartier et al., 2009). Activation of NMDA receptor upregulates UCHL1 activity and the levels of free ubiquitin monomers. Inhibition of UCHL1 disrupts distribution of synaptic proteins, increases dendritic spine size and reduces spine density. However, restoration of ubiquitin in UCHL1-inhibited neurons rescues impaired synaptic structure (Cartier et al., 2009). UCHL1 is also essential for normal structure and function of neuromuscular junctions (NMJs), reduction of synaptic vesicles and accumulation of tubulovesicular structures at the presynaptic nerve terminals have been observed in Uchl1 knockout mice (Chen et al., 2010). UCHL3, a homolog of UCHL1, is involved in spatial and working memory (Wood et al., 2005).

19S proteasome-associated DUBs comprise three members: USP14, UCH37, and RPN11, which show differences in ubiquitin chains removals and consequent protein degradation (Lee et al., 2011). It has been reported that both USP14 and UCH37 prevent substrate degradation by removing ubiquitin chains and promoting proteasomal substrate dissociation. In contrast, RPN11 cleaves at the base of the ubiquitin-linked substrates and promotes substrate degradation (Lee et al., 2011). USP14 is essential for synaptic development and normal function of NMJs (Chen et al., 2009), which is mediated through ubiquitination and activation of c-Jun N-terminal kinase signaling (Vaden et al., 2015). Pharmacological inhibition of USP14 enhances proteasomal degradation of several neurotoxic proteins, including tau, TDP14, and Ataxin-3 (Lee et al., 2010). Usp14-deficient mice display motor neuron deficits and developmental NMJ defects (Chen et al., 2009). USP16, a chromosome 21-encoded gene, is involved in chromatin remodeling and cell cycle progression through regulated ubiquitination of histone H2A (Joo et al., 2007). USP16 interacts with HERC2 and modulates the ubiquitination in DNA repair machinery components (Zhang et al., 2014). USP16 is upregulated in Down’s syndrome (DS) cells due to extrachromosomal triplication in trisomy 21, downregulation of USP16 partially restores the impaired proliferation in DS somatic stem cells (Adorno et al., 2013). It has been reported that USP16 acts as a H2A DUB and regulates hematopoiesis and hematopoietic stem cell function (Gu et al., 2016). Moreover, USP16 is required for embryonic stem cell differentiation, where USP16 deficiency leads to embryonic lethality in mouse deletion models (Yang et al., 2014).

The gene encoding USP25 also resides on human chromosome 21 (Valero et al., 1999), USP25 is involved in protein degradation by the 26S proteasome (Valero et al., 1999). USP25 is also implicated in inflammation, and it has been found to facilitate removal of Lys63-linked ubiquitin chains from tumor necrosis factor-receptor associated factor 5/6 (TRAF5/6), Usp25-deficient mice show higher sensitivity to IL-17 dependent inflammation and autoimmunity (Zhong et al., 2012).

Accumulating evidence indicates that dysfunction of the UPS is a key factor to initiate and aggravate the pathogenesis of neurodegenerative diseases. We will focus on the dysregulation of ubiquitination and proteasomal degradation in neurodegenerative pathology in AD, PD, and HD (Table 1).

Although UPS-mediated protein degradation has been well-studied in the last few decades, regulation of ubiquitination and proteasomal degradation in normal brain function and pathogenesis of neurodegenerative diseases remains largely unknown. The regulation of polyubiquitination processes may vary at different physiological and/or pathologic conditions. This review surveyed dysregulation of UPS in several major neurodegenerative diseases, more research is required to understand how the UPS is disrupted in neurodegenerative pathogenesis. By enriching our understanding of protein degradation, we can begin to formulate strategies to treat neurodegenerative diseases remediating aspects of UPS dysregulation.

QZ, LZ, and XW wrote the manuscript, TH, YZ, HL, and HX edited the manuscript. All authors read and approved the final manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.