Through tagging of proteins with ubiquitin to induce degradation, the UPS helps with maintaining protein homeostasis. This process begins with the activation of ubiquitin via ATP conversion to AMP and resultant attachment of ubiquitin to an E1 ligase, which is a ubiquitin-activating enzyme (Figure 1). The E1 ligase then transfers the ubiquitin to an E2 ligase, which is a ubiquitin-conjugating enzyme, and binds to a specific E3 ligase, generating an E2-E3-Ub complex. This E2-E3-Ub complex is primed for transfer of ubiquitin to a specific protein that binds to a particular E3 ligase. If the target protein is tagged via K48-linked ubiquitin, it is brought to the 26S proteasome and degraded. This process is essential for maintaining protein homeostasis and an integral part of normal cell function and survival (Jana, 2012).

E3 ubiquitin ligases are integral members of the UPS. There are three different E3 ubiquitin ligase classes, including really interesting new gene (RING), homologous to E6AP C-terminus (HECT), and ring-in-between-ring (RBR) E3 ligases, which are further subdivided into different sub-classes (Morreale and Walden, 2016). As shown in Figure 1, the RING E3 ligase ubiquitin transfer mechanism is arguably the simplest and involves ubiquitin transfer directly from the E3-E2-Ub complex to a lysine residue on the bound substrate. There are approximately 600 RING E3 ligases in humans, which are subdivided into monomeric, homodimeric, heterodimeric, monomeric U-box, homodimeric U-box, and cullin-RING types. The HECT and RBR E3 ligases ubiquitin transfer mechanism requires a primary transthiolation from a cysteine on the E2 ligase to a cysteine on the E3 ligase domain prior to transfer to the substrate (Figure 1). There are approximately 30 HECT E3 ligases in humans, divided into the neuronal precursor cell expressed developmentally downregulated 4 (NEDD4) and HECT and RLD domain containing (HERC) families, primarily. Finally, there are only approximately 12 RBR E3 ligases, including proteins such as Parkin, making it the smallest sub-class.

While the UPS is active during normal neurodevelopment, aberrant functioning of E3 ligases as part of the UPS has been suggested to drive neurodegeneration (Upadhyay et al., 2017). The ability of small molecule ligands to bind to E3 ligases provides a promising strategy to elicit either inhibition or activation and correct the mis-regulation that can occur (Gong et al., 2016). When the function of the UPS is impaired, degradation of either misfolded or damaged proteins may not occur, leading to aberrant cellular functions (Jana, 2012). Specifically in the case of AD, the UPS dysfunction plays a key role in the regulation of synaptic plasticity and transmission, regulation of mitochondria, autophagic flux, and endoplasmic reticulum (ER) homeostasis (Tydlacka et al., 2010).

As essential proteins involved in neuronal homeostasis and signaling are disrupted when the UPS is dysfunctional in AD, aberrant synaptic activities and pathways that result in neuronal death are propagated in AD patients. Synaptic function is often reduced in AD due to accumulation of misfolded and hyperphosphorylated-tau and this tau pathology results in neuronal apoptosis (Colom-Cadena et al., 2020). Zhou et al. (2017) demonstrated in rat and fly neurons that pathogenic frontotemporal dementia and parkinsonism can be linked to chromosome 17 (FTDP-17) mutant tau. FTDP-17 mutant tau forms microtubules in the axons of neurons, is localized to presynaptic sites, binds to synaptic vesicles, and impairs both synaptic vesicle mobility and release rate (Zhou et al., 2017). Soluble Aβ peptides and Aβ oligomers (AβO) that bind to synaptic receptors are linked to impairment of synaptic plasticity (Viola and Klein, 2015). Additionally, hyperactive microglial and neuroinflammatory mechanisms also play a role in synaptic dysfunction. Activated microglia can cause synaptic loss and release pro-inflammatory kinases, resulting in synaptic toxicity (Jackson et al., 2019). E3 ligases that are implicated in synaptic loss and/or dysfunction include ubiquitin protein ligase E3A (Ube3A) and NEDD4-1 (Rodrigues et al., 2016; Olabarria et al., 2019).

Likewise, impaired mitochondrial function in AD is made worse by E3 ligases that are not functioning properly. Based on analysis of post-mortem human brain tissue from the dorsolateral prefrontal cortex (BA41/42) of individuals with AD pathology, it was reported that AD results in reduced mitochondrial expression at presynaptic terminals within this brain region (Pickett et al., 2018). Furthermore, primary hippocampal neurons treated with AβO demonstrated less axonal trafficking of mitochondria toward synapses in an N-methyl-D-aspartate (NMDA) receptor-dependent manner (Decker et al., 2010). Damaged mitochondria are degraded via the PTEN induced kinase 1 (PINK1)/Parkin E3 ubiquitin ligase mediated pathways, by which mitochondrial outer membranes are ubiquitinated for mitophagic degradation (Wang W. et al., 2020). Mitochondria targeted by Parkin are believed to accumulate in the somatodendritic region of neurons and await lysosomal degradation (Cai and Jeong, 2020). Parkin-mediated mitophagy is activated by Aβ accumulation in mutant human amyloid precursor protein (hAPP)-expressing transgenic neurons and in AD patient brains, and a disease-related decrease in cytosolic Parkin results in mitophagy dysfunction (Ye et al., 2015).

E3 ligases also play roles in autophagy that are disrupted in AD. Among other roles, autophagy is a key process that mediates the clearance of misfolded and/or aggregated proteins (Uddin et al., 2018). Bordi et al. (2016) evaluated CA1 hippocampal pyramidal neurons from early and late-stage AD patients and reported reduced autophagic flux. This finding was based on the observations of poor substrate clearance, accumulation of promoters of autophagic degradation such as LC3-II and SQSTM1/p62 in autolysosomes, and changes in autolysosomal size and area. Long et al. (2020) demonstrated that in AD patient brains as well as in animal and cell models, Aβ can activate autophagy and mutation of the APP gene to APPswe (APP695 Swedish mutation) disrupts the fusion of autophagosomes with lysosomes, leading to impaired autophagy. E3 ligases that are implicated in autophagy include, but are not limited to Parkin, TNF receptor associated factor 6 (TRAF6), and NEDD4-1 (Kuang et al., 2013).

A final role of E3 ligases is related to the ER, which does not function properly in AD. One of the functions of the ER is the removal of misfolded proteins. ER stress in AD is linked to inflammatory pathways that interrupt this essential ER function, allowing protein aggregates to accumulate (Salminen et al., 2009). ER stress markers have been characterized in neurons and not glial cells, and prolonged ER stress can lead to synaptic loss and axonal degradation (Salminen et al., 2009). The E3 ligase HMG-CoA Reductase Degradation 1 Homolog (HRD1) is expressed in the ER membrane of brain neurons and is responsible for ER-associated degradation (ERAD), a process that will be discussed in more detail in a subsequent section devoted to HRD1 (Zheng et al., 2016).

E3 ligase ligands are important in the design and preparation of PROTACs, which are heterobifunctional molecules containing a ligand for an E3 ligase and a protein of interest (POI) that is targeted for proteasomal degradation. The PROTAC brings the E3 ligase into close proximity with the POI, forms a ternary complex, ubiquitinates a lysine residue on the POI, and elicits proteasomal degradation of the POI. Despite more than 600 known E3 ligases, only approximately 10 small molecule E3 ligase ligands of varying quality have been characterized (Ishida and Ciulli, 2021). Within the PROTAC field only two of these E3 ligases are routinely targeted, Von-Hippel Lindau (VHL) and cereblon (CRBN), since small molecule ligands have been identified (Fischer et al., 2014; Galdeano et al., 2014). Discovery of ligands for additional E3 ligases that can be recruited to ubiquitinate target proteins will enable development of PROTACs that induce protein degradation. Additionally, tissue- or disease-specific E3 ligases that are upregulated in AD patient brains could be targeted to localize the degradation. Furthermore, development of E3 ligase ligands with improved physicochemical properties could increase PROTAC cell permeability and possibly enable delivery to the brain.

The structural differences and complexities of E3 ligases preclude a one-size-fits-all approach for small molecule inhibitor or activator development. Additionally, the current paucity of chemical modulators available for E3 ligases includes several with poor physicochemical properties, suboptimal potency, and often unexplored selectivity. Development of chemical probe quality molecules is of high importance to enable unambiguous characterization of the role(s) that these proteins play in a cellular context (Frye, 2010; Arrowsmith et al., 2015). Herein, we review the specific E3 ligases implicated in AD with altered expression contributing to disease progression. We next provide an overview of the structural data available for these E3 ligases and published exogenous ligands. This information will be invaluable in any campaign directed at targeting E3 ligases as a method of therapeutic intervention for AD. To fill the gaps in this knowledge, several current strategies that are needed to pursue E3 ligase ligand discovery are summarized. Finally, the promising potential of the use of PROTACs for the treatment of AD is introduced as a future direction that is enabled by the knowledge and technologies highlighted in other chapters of this review.

Either upregulation or overactivity has been observed with respect to certain E3 ligases in AD patients (Figure 2 and Table 1). The overactivation of E3 ligases can occur via a variety of complex mechanisms dependent on the type of E3 ligase. These mechanisms include gain-of-function mutations, phosphorylation-induced autoubiquitination, and binding of adaptor proteins, among others (Buetow and Huang, 2016). Increased activity of an E3 ligase results in enhanced degradation of native targets. Examples of E3 ligases that are marked by unusually elevated expression or activity in AD include NEDD4-1, Itchy homolog (Itch), TRAF6, and ring finger protein 182 (RNF182), and membrane-associated RING-CH (MARCH8). It is suggested that small molecules could be used to inhibit the function or to degrade these aberrant E3 ligases with potentially therapeutically beneficial results for AD patients.

Decreased expression and/or reduced activity of certain E3 ligases has been observed in AD patients (Figure 2 and Table 1). E3 ligases that are reported as downregulated in AD include carboxy-terminus of Hsc70 interacting protein (CHIP), tetratricopeptide repeat protein 3 (TTC3), Ube3A, HRD1, and Parkin. Either rescuing or enhancing the activity of these E3 ligases where mutation and/or reduced expression has resulted in loss-of-function may be beneficial for AD patients.

Elucidation of the structure of E3 ligases via crystallographic or NMR-based methods greatly facilitate drug discovery efforts. Currently, a number of E3 ligases that have been linked to AD pathology, several of which were described in the previous section, have been crystallized and have solved structures: CHIP, NEDD4-1, TRAF6, Parkin, HRD1, Ube3A (E6AP), Itch, and MARCH8 (Table 1). Additionally, co-crystallographic structures of a few E3 ligases with peptides or small molecule ligands bound have been disclosed. The availability of co-crystal structures is pivotal and creates the potential for more expedient ligand design. While not essential for small molecule inhibitor development, an available structure enables computational studies and structure-based drug design as integral parts of the ligand development process. CHIP has been co-crystallized with its corresponding native peptides and an optimized peptide ligand (Figure 3). NEDD4-1 is unique as it has published co-crystal structures with both peptides and covalent small molecule inhibitors (Figure 3). To our knowledge, there is no crystallographic data for E3 ligases TTC3 and RNF182. We highlight a few of these structures and what has been learned through their public availability in the section below.

Ligands that bind to specific E3 ligases in a selective manner will be impactful in elucidating mechanisms and the therapeutic potential of these putative targets within AD. Ligands that can activate the function of an E3 ligase that is downregulated and/or has loss-of-function mutations could aid restoration of biological function(s) that may slow AD progression. Alternatively, E3 ligases can be inhibited via different mechanisms: (1) by impeding E2 ligase binding to the E3 ligase, (2) by disrupting substrate binding of the E3 ligase, or (3) for HECT and RBR E3 ligases that require an intermediatory transthiolation of ubiquitin, by blocking the transfer of ubiquitin onto the E3 ligase (Figure 4) (Spratt et al., 2014; Weber et al., 2019). E3 ligands can be either small molecules or peptides, and examples of each have been described. The development of small molecule ligands that have favorable physicochemical and pharmacokinetic/pharmacodynamic properties is desirable, and ultimately optimization of a ligand for blood–brain barrier penetration is necessary for AD-related targets in the brain. Peptidic ligands are often designed based on the shortest peptide sequence found within the binding domain of endogenous binding partners. By nature, these peptidic ligands are not drug-like but can provide valuable starting points for small molecule ligand development. One final category of E3 ligase-binding ligands includes bifunctional molecules, such as PROTACs and deubiquitinase-targeting chimeras (DUBTACs), which bring E3 ligases in proximity with another protein to promote either degradation or stabilization. The development of PROTACs that can induce degradation of previously undruggable targets using either small molecule or peptide ligands has provided an alternative approach to modulation of protein function (Paiva and Crews, 2019). PROTACs demonstrate different properties than small molecule inhibitors and act via an event-driven pharmacology as opposed to an occupancy-driven pharmacology (Lai and Crews, 2017). Unique characteristics of the event-driven pharmacology elicited by PROTACs includes their catalytic activity and ability to remove function until a protein is resynthesized. Additionally, the specificity of degradation of the protein of interest (POI) is not governed by the specificity of the parent ligand, but by the ternary complex interface formed between the E3-PROTAC-POI (Lai and Crews, 2017; Smith et al., 2019). Recently, the development of DUBTACs has furnished bifunctional molecules that are capable of stabilizing proteins. DUBTACs have utility in diseases like AD when the ubiquitination and degradation of key proteins drives disease pathology. These compounds provide a way to stabilize the actively degraded target and therein preserve its essential function, slowing disease progression (Henning et al., 2021). Examples of known ligands for AD-relevant E3 ligases are included in the following subsections (Table 1). Although the ligands listed below are beneficial for beginning to understand the effects of modulating the function of these E3 ligases, it is important to note that none of these ligands fit in the criteria to be considered a chemical probe, highlighting the need for additional research effort. A chemical probe has sufficient on-target potency, selectivity, a defined mechanism of action, full characterization to enable in vitro use to address a specific biological hypothesis, and must be widely available to the scientific community (Frye, 2010).

Multiple assays have been developed to identify ways to inhibit E3 ligases that are overactive or overexpressed in AD. The function of E3 ligases can be inhibited via impeding ubiquitin transfer from the E2 ligase to the E3 ligase or inhibiting the E2 ligase from interacting with its E3 ligase (Figure 4). Direct inhibition of E3 ligase activity can be achieved via targeting and/or modifying the cysteine responsible for transthiolation of ubiquitin in the case of RBR and HECT family E3 ligases (Figure 4). Inhibition of E3 ligase function can be determined via mass spectrometry (MS), luminescence- and/or fluorescence-based methods.

The activity of E2 and E3 ligases can be determined via measuring the depletion of mono-ubiquitin using a high-throughput matrix-assisted laser desorption/ionization-time of flight (MALDI-TOF) assay (Cesare et al., 2018). An assay of this type is built on knowledge that in the absence of a substrate, an E3 ligase will either produce polyubiquitin chains or undergo autoubiquitylation, which can be monitored via ¹⁵N-labeled ubiquitin internal standards to observe the decrease in monoubiquitin via MALDI-TOF MS (Cesare et al., 2018). This method facilitates identification of specific E2/E3 substrate pairings that can be further used to develop a high-throughput assay specific to an E2/E3 ligase pair in pursuit of an inhibitor of their activity. Cesare et al. (2018) screened for inhibitors of MDM2 (a RING E3 ligase), Itch (a HECT E3 ligase), and HOIP (RNF31, a RBR E3 ligase) using a library of 1,430 compounds. Although this assay does not distinguish between inhibitors of E1/E2/E3 ligases, it pinpoints loss of activity of these pairings. MALDI-TOF MS can then be used to delineate specific interactions between the inhibitor and the individual E1/E2/E3 ligases. Cesare et al. (2018) were able to identify inhibitors of the E1-E2-Itch complex, however, they were found to be inhibitors of E2 ligase activity. Strategies to block E3 ligase activity by targeting cysteine residues responsible for ubiquitin transthiolation on RBR and HECT E3 ligases have been used to identify covalent cysteine modifiers (Johansson et al., 2019).

Electrochemiluminescence-based assays that require activation of the E3 ligase with E1 and E2 ligases plus ATP can be utilized for high-throughput screening to identify E3 ligase inhibitors (Kenten et al., 2005). This methodology relies on adhering the GST-E3 ligase to a plate coated with glutathione electrodes, followed by addition of an E1-E2-Ub mixture and putative inhibitors. The anti-Ub antibody with an appended MSD sulfa-TAG binds to ubiquitin and enables detection of ubiquitination via electrochemiluminescence. This method was employed in a screen of 10,000 compounds in an attempt to identify an inhibitor of the E3 ligase RNF28.

A ubiquitin ligase profiling (ULP) system was developed that utilizes a high-throughput cellular luciferase reporter assay format to evaluate the inhibition of RING type E3 ligases (Maculins et al., 2016). The assay relies on a Gal4-Firefly luciferase (FL) reporter assay with a Gal4 DNA-binding domain fused to the E3 ligase of interest. Upon ubiquitination of the E3 ligase, Tandem Ubiquitin Affinity Entities (TUBEs) can bind to ubiquitin, leading to recruitment of a fused VP16 activation domain (ACT) to the minimal TATA box of the Gal4-firefly luciferase and a resultant luciferase signal. The ULP system was applied to screen for inhibitors of the E3 ligase RNF8 with a secondary screen for selectivity against TRAF6.

Due to a need to regulate E3 ligases in AD, either via inhibition or activation, assays that have a dual readout and can be used to identify either inhibitors or activators in a high-throughput format are very useful. Examples of these types of assays are described below.

Homologous to E6AP C-terminus and RBR E3 ligases require transthiolation of a ubiquitin onto a catalytic cysteine residue prior to transfer of ubiquitin to the substrate. HECT E3 ligases have the catalytic cysteine in the HECT domain on their C-terminus, while RBR E3 ligases have the cysteine residue present in the RING2 domain, which also conjugates to ubiquitin (Smit and Sixma, 2014; Weber et al., 2019). Traditionally assays aimed at probing E3 ligase activity require an E1 ligase, E2 ligase, and ATP. The development of probes attached to ubiquitin, however, allow efficient transfer of ubiquitin to the E3 ligase via transthiolation followed by ligation to a substrate without the need for these components (Park et al., 2017). This system that does not require E1 and E2 ligases or ATP is called a ‘bypassing system’ (ByS) (Park et al., 2014).

The first example of one of these probes, called a ubiquitin mercaptoethanesulfonate (UbMES), is a ubiquitin C-terminal thioester probe that contains a mercaptoethanesulfonate group (Park et al., 2014, 2017). The effectiveness of the interaction of the E3 ligase catalytic cysteine with the UbMES probe is analyzed via western blot analysis, probing the extent of ubiquitination of the substrate, and allowing qualitative visual detection of inhibition or activation of substrate ubiquitination. A fluorescent labeled cysteine (FCys) can be added to quantitatively assess the enzyme kinetics of the HECT E3 ligase in this system. Park et al. (2014, 2017) demonstrated that other E3 ligases can adopt the ByS, including HECT E3 ligases (NEDD4-1, NEDD4-2, WWP1, and Itch) and RBR E3 ligases (Parkin and HHARI). A similar assay was developed called Ubfluor that releases a FP signal after ubiquitin is transferred to the E3 ligase, and inhibition of this transfer via introduction of covalent inhibitors quenches the signal (Krist et al., 2016; Foote et al., 2017). This FP format allows for a more high-throughput and quantitative assay that is not reliant on western blot analysis or addition of FCys and can be used to directly screen for inhibition of the catalytic cysteine in real time.

Tian et al. (2019) used a ubiquitin reference technique (URT) in a dual luciferase assay where K48-linked ubiquitin is tagged with renilla luciferase (RL) and a substrate for the E3 ligase is tagged with firefly luciferase (FL). The activity of the E3 ligase is determined via examination of the FL/RL ratio (Tian et al., 2019). This ratio can be used to identify inhibitors of E3 ligase activity (increase in FL/RL ratio) as well as the presence of E3 ligase activators (decrease in FL/RL ratio).

The AD biology is complex, making it essential to understand the native interactions and substrates of E3 ligases in hopes of identifying therapeutic intervention modalities and effective small molecule ligands. Methods that enable identification of E3 ligase substrates include proteomics (to evaluate ubiquitination or PPIs), yeast two-hybrid (Y2H) assays, and in vitro phage display screening, among other techniques (Iconomou and Saunders, 2016).

Quantitative proteomics can be performed to identify the substrates of E3 ubiquitin ligases. Lee et al. (2011) reported the use of immunoaffinity enrichment of all ubiquitinated proteins from cells expressing His-tagged ubiquitin, followed by elution through an immobilized metal affinity chromatography (IMAC) chemical affinity column that is specific for His-tagged ubiquitin. The proteins were then separated via SDS-PAGE, and the bands digested and analyzed by liquid chromatography-mass spectrometry with tandem mass spectrometry (LC-MS/MS) (Lee et al., 2011). Determination of ubiquitination changes was carried out using stable isotope labeling by amino acids in cell culture (SILAC) methods, with the HA-ubiquitin negative control expressed in light medium and the His-ubiquitin in heavy SILAC conditions. Increased expression of ubiquitinated proteins in the heavy SILAC conditions were observed for substrates of the E3 ligase. As part of an orthogonal approach, Lee et al. (2011) also employed a three-way SILAC with a control siRNA compared to two HRD1 siRNAs, with the expectation that HRD1 substrates would have decreased ubiquitin levels. They observed peptides after a tryptic digestion step. Overall, approximately 400 ubiquitinated proteins and 1800 ubiquitinated peptides were identified for HRD1 using this strategy.

Martinez et al. (2018) highlight in their work a method to identify Parkin and Ube3A E3 ligase substrates using a ^bioUb (biotinylated ubiquitin) strategy in Drosophila neurons (Ramirez et al., 2018). Developing as well as mature neurons in Drosophila have been employed. This strategy relies on Drosophila engineered to express an Escherichia coli biotin ligase (BirA) that can biotinylate ubiquitin in photoreceptor neurons and result in incorporation of ^bioUb. Overexpression of the WT and mutant inactive negative control E3 ligases, such as Parkin and Ube3A, in the fly neurons is required. A pulldown of ^bioUb purified with neutravidin beads is followed by MS analysis that enables identification of substrates of the E3 ligases. While Ube3A was found to be active in both developing and mature Drosophila neurons, Parkin was only active in adult neurons. Ramirez et al. (2018) identified 79 substrates of Ube3A with increased ubiquitination upon overexpression of Ube3A in Drosophila neurons. Ube3A substrates included proteasomal proteins, Rpn10 and Uch-L5, as well as others that implicated a role of Ube3A in neuronal morphogenesis and synaptic transmission. Parkin substrates included proteins related to mitophagy, endocytic trafficking, ER stress, immunity, and apoptosis. Interestingly, the PINK1-mediated Parkin activation mechanism is not present in this model system, potentially indicating alternative ways to activate Parkin. Although these data suggest novel substrates for E3 ligases, follow-up studies in human neurons will be necessary.

More recently, Ordureau et al. (2018) used a quantitative proteomics approach called Parkin target-parallel reaction monitoring (Pt-PRM) to characterize Parkin substrate ubiquitination on the mitochondrial outer membrane (MOM) in neuronal cells. The Pt-PRM methodology allowed examination of kinetics, site specificity, and stoichiometry in parallel (Ordureau et al., 2018). Quantitative analysis of endogenous Parkin substrate ubiquitination in human embryonic stem cell (hESC)-derived NGN2-induced neurons (iNeurons) and DA neurons versus in HeLa cells was executed. In neurons expressing endogenous levels of Parkin, chain linkage analysis suggested little K48-linked chain production. This was not surprising given that K63-linked chains are predominantly linked to Parkin-mediated mitophagy. Finally, in neurons and HeLa cells no significant reduction in the 15 MOM Parkin substrates used in this study was observed at either a 1 or 4 h timepoint after depolarization.

An engineered protein technique using ubiquitination-activation interactions traps (UBAIT) employ a ubiquitin fused to a particular E3 ligase to detect the substrates of that E3 ligase (O’Connor et al., 2015; Zhao et al., 2020). The substrate binds to the target and forms an amide-linked E3-Ub-target complex that can be identified via proteomics analysis. This approach has proven applicable to HECT and RING E3 ligases, including the HECT E3 ligase Itch. A benefit of this approach is that it can capture relatively weak interactions.

Yeast two hybrid approaches can be used to identify PPIs in live yeast cells. A transcriptional factor such as Gal4 and a reporter gene are expressed, and the target protein is attached to a DNA binding domain that binds to Gal4 (Iconomou and Saunders, 2016). The putative binding partner is attached to a transcriptional activation domain so that if the target protein and the binding partner interact, then transcription of the reporter occurs. This approach was used with NEDD4 via cloning NEDD4 cDNA to a Gal4 binding domain and a cDNA library cloned with the Gal4 activation domain pGAD10 (Murillas et al., 2002). N4BP1 was identified as a monoubiquitylation nuclear protein, while NFBP3 was found to co-localize with NEDD4 in cytoplasmic vesicles. Alternatively, orthogonal ubiquitin transfer (OUT) techniques in yeast also enable identification of substrates via engineering the yeast cell surface display so that an E3 ligase can only transfer affinity-tagged ubiquitin variants (Wang et al., 2017). Wang et al. (2017) applied this approach to Ube3A and determined substrates including MAPK1, CDK1, CDK4, PRMT5, UbxD8, and β-catenin.

As a methodology to enable identification of E3 ligase ligands, phage display is a biological strategy in which UbVs are generated that bind to E3 ligase Ub-binding domains or catalytic domains. These UbVs can inhibit HECT, monomeric RING/Ubox, or multi-subunit SCF RING E3 ligases or deubiquitinases. Alternatively, they can activate HECT or homodimeric RING E3 ligases (Ernst et al., 2013; Liu et al., 2021). In the case of RING/Ubox E3 ligases, inhibition occurs via blocking the E3-ubiquitin binding surface and activation occurs via stabilization of the E2-ubiquitin interaction (Gabrielsen et al., 2017). Zhang C. W. et al. (2016) demonstrated that with HECT E3 ligases, the UbV-based inhibition is due to blocking the E2 binding site, whereas activation is observed in response to UbV binding to the N-lobe exosite. Activation is suggested to be the result of release from an autoinhibited state (Zhang C. W. et al., 2016). Ordureau et al. (2018) developed a UbV that selectively inhibits Parkin^295–465 and Parkin^410–465 from binding to HECT E3 ligases (EDD1, HUWE1, WWP2, and HERC2) as well as non-specific controls (GST and BSA). The development of UbVs for phage display that can bind to domains of E3 ligases to elicit either inhibition or activation of function is a promising step in the pursuit small molecule ligands that would have potential utility in AD.

To screen for E3 ligase substrates, in a similar fashion to yeast cell surface display techniques, phage display techniques such as biopanning rely on the binding affinity of a protein library to immobilized receptors (Zhao et al., 2020). A protein library is fused to a protein (pIII) that is on the phage and, upon binding of the protein to immobilized receptors, a high-affinity phage-protein pair can then be selected and isolated. Bhuripanyo et al. (2018) used phage display techniques to engineer E2-E3 pairs compatible with the previously described OUT technique in which ubiquitin cascades only incorporate mutant ubiquitin. This strategy enabled identification of CHIP substrates, such as β-catenin and CDK4 (Bhuripanyo et al., 2018). Bhuripanyo et al. (2018) suggest that ER stress leads to the degradation of CDK4 via ubiquitination and proteasomal degradation, which could be mediated by CHIP.

The growing number of protein engineering and proteomics techniques should make the elucidation of the substrates of E3 ligases implicated in AD feasible. In turn, a better understanding of the complex biological functions of these E3 ligases may aid the development of novel chemical entities to modulate their function(s).

Proteolysis targeting chimeras (PROTACs) are heterobifunctional molecules that contain a ligand for an E3 ligase and a ligand for a POI attached via a linker region (Pettersson and Crews, 2019). The PROTAC brings the E3 ligase into close proximity to the POI via formation of a ternary complex, which facilitates the transfer of a ubiquitin from the Ub-E2-E3 complex onto a lysine residue on the POI. Attachment of ubiquitin to the POI as K48-linked polyubiquitin chains induces 20S proteasomal degradation of the POI and releases the PROTAC to degrade more POI. This catalytic mechanism of degradation of the POI means that lower and/or less frequent doses of PROTACs are often needed (Mares et al., 2020). In addition, the effect of degradation as opposed to inhibition can often lead to enhanced or longer-lived cellular effects when dosing with PROTACs (Wang and Zhou, 2018; You et al., 2020). PROTACs have recently proven effective in the clinic with androgen and estrogen receptor targeting PROTACs for cancer (Mullard, 2019).

Despite the success of PROTACs, there is still only a paucity of ligands available for E3 ligases (Ishida and Ciulli, 2021). Additional opportunities for the development of ligands for E3 ligases include tissue-specific E3 ligases that could elicit targeted degradation effects within the body. The development of novel ligands for E3 ligases presents the opportunity to develop PROTACs that degrade undruggable proteins within the proteome that are not currently targeted via available E3 ligases. Furthermore, new E3 ligase ligands could provide variable degradation potential of highly homologous proteins within the same family, offering enhanced selectivity versus traditional inhibition (Smith et al., 2019).

Current approaches toward PROTAC-mediated degradation within AD are focused on the targeted degradation of tau and hydrophobic tagging of tau that does not rely on E3 ligase recruitment (Gao et al., 2017b; Hyun and Shin, 2021). The first reported tau-degrader, TH006, by Chu et al. (2016) was a peptidic-based PROTAC with a cell penetrating peptide motif attached. TH006 reduced tau levels and improved viability in Aβ-treated primary neuron cells and in 3xTg-AD mice, which harbor three mutations associated with familial AD and exhibit both tau and Aβ pathology. In a similar approach, Lu et al. (2018) developed a tau-targeting PROTAC that contained a peptidic ligand for the E3 ligase Keap-1. This PROTAC induced tau degradation in SH-SY5Y, Neuro-2a, and PC-12 tau-overexpressing cell lines. Silva et al. (2019) adapted this strategy to develop a PROTAC, QC-01-175, that is comprised of a positron emission tomography (PET) probe appended to a cereblon (CRBN) E3 ligase ligand via a PEG2 linker. QC-01-175 induced a reduction in levels of tau and hyperphosphorylated tau (p-tau S396, S404) in human neuronal cell models derived from PSP patients with a tau-A152T risk variant and behavioral-variant FTD patients with a tau-P301L autosomal dominant heterozygous mutation. Additionally, QC-01-175 can differentiate between the disease-associated tau and normal tau in iPSC-derived neuronal cell models. Global proteomics analysis following 4 h treatment with 1 μM of QC-01-175 in A152T neurons demonstrated off-target degradation of the proteins ZFP91, ZNF653, and ZNF827, which are known immune-modulatory drug targets of CRBN. Finally, Wang et al. (2021) reported a tau degrader, C00419, that contains a VHL E3 ligase recruiter (VH032-derived ligand) and a tau ligand connected via a PEG3 linker region. C00419 was able to clear tau in vitro in SH-SY5Y cells as well as in HEK293 cells either stably or transiently transfected with hTau and in vivo in wild-type, 3xTg and hTau mice (Wang et al., 2021). Spine density was measured in neurons from 3xTg mice following subcutaneous administration of multiple doses of C004019 (3 mg/kg), which resulted in an increase in spine density. Similarly, in hTau mice the spine density and neurite arborization in the hippocampal dentate gyrus increased following subcutaneous administration of multiple doses of C004019 (3 mg/kg). Structures of these promising tau-targeting PROTACs are featured in Figure 5.

The development of tau-targeted PROTACs has pioneered the idea that PROTACs can directly induce degradation of key components that drive AD pathology. The generation of Aβ peptides from the cleavage of APP and accumulation of Aβ in AD is a classic hallmark of the disease. Despite knowledge of the amyloidogenic pathway, clearance of Aβ has proven a challenging therapeutic direction and there is not currently a small molecule strategy that allows direct binding and modulation of Aβ. As PROTACs are being used to target ‘undruggable proteins,’ eliciting Aβ degradation via a PROTAC offers a way to clear these pathogenic accumulations and could represent a viable strategy in AD (Gao et al., 2017a). The availability of Aβ PET ligands could enable a similar strategy to that employed by Silva et al. (2019) with respect to tau, in which Aβ PET ligands could be converted to Aβ-targeting PROTACs. Additionally, PROTACs targeting overexpressed or overactive proteins implicated in the pathogenesis of AD for degradation could lead to the regulation of novel pathways within AD. DUBTACs, conceptually similar to PROTACs in that both approaches utilize heterobifunctional molecules, is an emerging strategy that leads to deubiquitination of proteins, and has potential to further enable new therapeutic directions in AD (Henning et al., 2021). DUBTACs offer the ability to stabilize proteins that are otherwise downregulated in AD, thus providing new and different disease modifying strategies versus inhibition or degradation. PROTAC development within neurodegeneration is dependent on the generation of ligands for the desired target for degradation, and generation of ligands for the E3 ligases to be harnessed. Although tissue-specific and/or disease specific E3 ligases that could be targeted via PROTACs have been identified, the lack of available small molecule ligands impedes their development and evaluation of these therapeutic hypotheses. A paucity of small molecule ligands is available for the dysfunctional AD E3 ligases mentioned herein this review. A key first step toward enhanced understanding of the underlying disease pathology of E3 ligases in AD is the design of high-quality chemical probes to assess the effects of either inhibition, activation, or degradation of these E3 ligases on the key phenotypes of AD in neurons.

The dysfunctional role of UPS also drives the pathogenesis and regulation of aggregated proteins in neurodegenerative diseases such as amyotrophic lateral sclerosis (ALS) and Parkinson’s disease (PD). Since UPS dysfunction is exacerbated by protein aggregation, it contributes to the neuronal dysfunction and/or loss that plagues patients. Specifically, in PD the accumulation of α-synuclein (α-syn) generates Lewy Bodies that can hamper synaptic function via impaired UPS, reduced autophagy, and damaged lysosomes (Guerroué and Youle, 2021). Within ALS, protein aggregates of TAR DNA-binding proteins 43 (TPD-43), superoxide dismutase type-1 (SOD1), and RNA-binding protein fused in sarcoma (FUS) are linked to mitochondrial dysfunction and degeneration, impaired UPS, and reduced autophagy (Chisholm et al., 2021). Farrawell et al. (2020) identified UPS genes that are either up or downregulated in ALS neurons, noting a predominant decrease in the expression of UPS genes in ALS. Despite this, several E3 ligases are upregulated in ALS-afflicted neurons. Some of the same E3 ligases discussed with respect to AD that could represent therapeutic targets for AD also play essential roles in the pathology of PD and/or ALS. Known exogenous ligands that were included in the AD-specific sections may find utility in treating these other neurodegenerative diseases.

Aggregated proteins are a problem common to several neurodegenerative diseases, including AD, PD, and ALS. The use of PROTACs to target these misfolded and/or aggregated proteins for degradation is a viable therapeutic approach. As an example, tau-directed PROTACs as potential AD therapeutics were discussed in a previous section.

Several strategies have been employed to target α-syn for proteasomal and lysosomal degradation. Firstly, Fan et al. (2014) reported a synthesis of a peptide containing an α-syn-binding sequence, cell penetrating peptide (CPP) component, and an autophagy-targeting motif that induced α-syn degradation in primary cultured rat neurons. Although this strategy was effective in their cellular model, lysosomal impairment is a common hallmark of PD that is responsible for accumulation of α-syn and which may reduce the efficacy of this and related approaches (Navarro-Romero et al., 2020). Qu et al. (2020) designed a peptide-based PROTAC with a TAT-protein binding domain-proteasome targeting motif (TAT-PBD-PTM), containing a TAT CPP that simultaneously binds to intracellular α-syn and E3 ligases to induce α-syn proteasomal degradation. The PTM is comprised of a small tetrapeptide with the amino acid sequence RRRG, which was previously identified as a degron that induces proteasomal degradation (Bonger et al., 2011). Via mass spectrometry Qu et al. (2020) identified that this peptide region binds to E3 ligases TRIP12, STUB1, and UHRF1. Furthermore, significant degradation of α-syn was observed in SK-N-SH cells and in primary cortical neurons following treatment with TAT-PBD-PTM (25 μM). This peptide-based PROTAC induced a decrease in endogenous α-syn by >70% in primary neurons that were cultured for 8 days with treatment of TAT-PBD-PTM at 25 μM for 4 h. Finally, the PROTAC-based company Arvinas has patented PROTACs targeting α-syn that employ small molecule E3 ligase ligands for VHL and CRBN appended to α-syn-targeting compounds. These PROTACs result in <35% α-syn after treatment at 1 μM for 48 h in HEK293 TREX α-syn A53T cells (Kargbo, 2020).

To our knowledge, there are currently no reported ALS relevant PROTACs, which could be used to target protein (mutant SOD-1 or TDP-43, for example) aggregates found in ALS patients. Rao et al. (2021) summarized the current landscape of TDP-43 ligands for the different domains of TDP-43. Solved crystal structures of TDP-43 have enabled the development of TDP-43 ligands. These ligands consist of small molecules, peptides, and oligonucleotides and could be applied to the development of PROTACs akin to the tau-targeted degraders that utilize known PET tracers for tau. Ligands of SOD-1 have not been described in the literature, precluding SOD1-targeting PROTAC development at this point.

The development of novel E3 ligase ligands that have disease-specific functions or tissue-specific locations could greatly enable the development of PROTACs aimed at treating certain neurodegenerative diseases. The initial success of PROTAC technology for the degradation of aggregated and/or misfolded proteins in neurodegenerative diseases demonstrates that further development in this area could be therapeutically beneficial in diseases such as AD, PD, and ALS. This is a promising future direction of the PROTAC and neurodegenerative fields.