Proteolysis Targeting Chimeras as Therapeutics and Tools for Biological Discovery

Abstract

New biological tools provide new techniques to probe fundamental biological processes. Here we describe the burgeoning field of Proteolysis Targeting Chimera (PROTACs) which are capable of modulating protein concentrations at a post-translational level by co-opting the Ubiquitin-Proteasome System. We describe the PROTAC technology, its application to drug discovery and provide examples where PROTACs have enabled novel biological insights. Furthermore, we provide a workflow for PROTAC development and use as well as a discussion of the benefits and issues associated with PROTACs. Finally, we compare PROTAC mediated protein level modulation with other technologies such as RNA interference and genome editing.

Introduction

Technological advances often lead to major biological discoveries (Botstein, 2010; Editorial, 2000; Fields, 2001; van Steensel, 2015), which, in turn, drive new technology development. The search for new technologies to answer biological questions has given rise to the field of chemical biology (Altmann et al., 2009) and perhaps one of the most exciting technologies to arise hence is targeted protein degradation by Proteolysis Targeting Chimeras (PROTACs). In this primer, we will review this technology, its applications to drug discovery and its use in the exploration of fundamental biology. Additionally, we will outline the workflow involved in successfully employing PROTACs for experimentation and compare PROTACs with other techniques.

Overview of the Technology

The Ubiquitin-Proteasome System

The Ubiquitin-Proteasome System (UPS) is the primary intracellular mechanism for the destruction of damaged proteins or those no longer required (Amm et al., 2014). This has been extensively reviewed elsewhere (Kleiger and Mayor, 2014), however, a brief explanation of the UPS is provided here as it pertains to PROTACs. The 76 residue protein ubiquitin is attached to proteins via a lysine isopeptide bond as a post-translational modification (PTM) via a cascade of three enzymes; an E1 activating enzyme, an E2 conjugating enzyme and an E3 ligase (Figure 1). Free ubiquitin is activated by an E1 in an ATP-dependent process during which it is converted to a C-terminal thioester (Figure 1). Trans-thioesterification passes the ubiquitin from the E1 unto an E2. Finally, an E3 complex facilitates the transfer of ubiquitin, either directly or indirectly, to a substrate protein lysine.

Figure 1 –

Schematic representation of the ubiquitin cycle and how it can be co-opted by PROTACs. Ubiquitin-E1 thioester (UBA1 Ub complex, PDB ID:3CMM). Ubiquitin-E2 Thioester (UBE2D3-UbDha Complex PDB ID: 5IFR). Ubiquitin-E2-E3-Substrate Complex (Model Composed of PDB ID:1LQB, 5N4W and 5IFR). Ubiquitin-E2-E3-PROTAC-Neosubstrate Complex (Model Composed of PDB ID:5T35, 5N4W and 5IFR). Proteasome (PDB ID: 5I4G)

Ubiquitin itself can, in turn, be ubiquitinated on one or more of its seven surface lysine residues. These PTMs have a multitude of biological functions that are still being investigated (Komander and Rape, 2012; Yau and Rape, 2016), but the canonical role of K48 polyubiquitin is to signal a protein for destruction via the proteasome (Grice and Nathan, 2016). Enzymes involved in protein ubiquitination are present in the human genome in increasing numbers as they progress from E1, of which there are only 2, to E3 with >600 postulated E3 family members (Ardley and Robinson, 2005). The E3 ligases serve to recruit substrates and facilitate the transfer of ubiquitin from an E2 conjugating enzyme to the target protein. Once a protein is tagged with poly-K48 ubiquitination and recognized by the proteasome, the ubiquitin chains are removed by proteasome associated deubiquitinating enzymes (DUBs) and recycled, while the protein substrate is unfolded and degraded (Figure 1).

Hijacking the UPS System

The PROTAC technology allows the UPS system to be chemically co-opted and aimed to degrade a specific target protein. This approach employs E3 ligase ligands, fused via a flexible chemical linker to a targeting element for the protein of interest, to elicit ectopic ubiquitination resulting in protein degradation (Figure 1). Beginning with proof-of-concept experiments in cell lysates with peptidic ligands (Sakamoto et al., 2001), the technology has matured and is routinely used in cultured cells, in vivo and has even entered clinical trials. Importantly, the technology is now comprised of all-synthetic, modular compounds (Bondeson et al., 2015; Winter et al., 2015), which function against a wide range of protein classes and in different subcellular locations including the cytosol, the nucleus and the plasma membrane (Burslem et al., 2018b). As we come to understand this new technology better, it is clear that protein-protein contacts between the neo-substrate and E3 complex are a crucial determinant of PROTAC success, as shown in Figure 2 (Bondeson et al., 2018; Gadd et al., 2017; Nowak et al., 2018; Smith et al., 2019; Zorba et al., 2018). The Ciulli group has been instrumental in the structural and biophysical characterization of co-operativity in this area (Farnaby et al., 2019; Roy et al., 2019).

Figure 2 -.

Composition of a PROTAC for BRD-4. Left – Structure of VHL ligand bound to VHL (PBD ID: 6FMI). Center – PROTAC (MZ-1)-induced ternary complex between VHL and BRD-4 (PDB ID: 5T35). Right – Structure of JQ-1 bound to BRD-4 (PDB ID:3MXF)

Only a handful of E3 ligases have been employed in the PROTAC technology (Table 1), with most of the reported compounds using either CRBN or VHL due to the availability of drug-like small molecules that recruit them (Buckley et al., 2012; Winter et al., 2015). cIAP ligands have also been used but often lead to auto-ubiquitination and degradation of the E3 ligase itself, making them less attractive (Sekine et al., 2008). The use of nutlin compounds to recruit MDM2 played a key role in the development of the first all-small molecule PROTAC (Schneekloth et al., 2008) and has recently reemerged as a potent method for targeted protein degradation (Hines et al., 2019). With >600 postulated members of the E3 superfamily, it is exciting to watch ligands emerge for new E3 ligases and their application to targeted protein degradation (Ottis et al., 2017; Spradlin et al., 2019). Indeed, some of these new ligands allow degradation only in specific cellular compartments due to their restricted localization (Zhang et al., 2019b). Given the importance of E3/target protein-protein interactions in PROTAC-mediated degradation, the more E3s available for recruitment provide a greater chance of successful PROTAC development.

Table 1 -.

E3 Ligases currently used in PROTACs

E3 Adaptor	Native Substrate	Ligand	Key Reference(s)
VHL	Hydroxylated HIF-1α	VHL Peptidomimetics	(Bondeson et al., 2015; Buckley et al., 2012; Schneekloth et al., 2004)
CRBN	Glutamine Synthetase MEIS2	IMiDs	(Lu et al., 2015; Winter et al., 2015)
MDM2	p53	Idasanutlin	(Hines et al., 2019; Schneekloth et al., 2008)
β-TRCP	β-Catenin	Phosphorylated Peptide	(Sakamoto et al., 2001)
cIAP	second mitochondria-derived activator of caspases (SMAC)	Bestatin Esters MV-1 LCL161 (SMAC Mimetics)	(Itoh et al., 2010)
RNF4	Poly-Sumoylated Proteins	CCW-16 (Covalent Fragment)	(Ward et al., 2019)
RNF14	p21	Nimbolide (Natural Product)	(Spradlin et al., 2019)
DCAF16	unknown	KB02 (Covalent Fragment)	(Zhang et al., 2019b)

Applications

PROTAC technology has applications in both biological discovery and drug discovery. In many ways, PROTACs represent the chemical equivalent of siRNA, albeit allowing for the removal of a protein at a post-translational level rather than silencing at a post-transcriptional level. As such, they represent a useful tool for studying the roles of a protein in biological systems in the laboratory. Additionally, the small molecule nature of PROTACs circumvents issues associated with delivery and biodistribution that hinder clinical applications of siRNA, thereby eliciting great interest from the pharmaceutical industry.

PROTACs in Drug Discovery

Unsurprisingly, given its small molecule nature, PROTAC technology is progressing into the clinic for numerous indications. Initial studies have focused on the degradation of hormone receptors (Flanagan and Neklesa, 2019), specifically androgen receptors (Neklesa et al., 2019) and estrogen receptors (Flanagan et al., 2019). Notably, these trials are for orally bioavailable PROTACs highlighting the benefits of PROTACs over therapeutic RNAi (Setten et al., 2019) or traditional selective estrogen down regulators (SERDs) (Patel and Bihani, 2018).

Androgen Receptor

Androgen receptor (AR) antagonists, such as enzalutamide (Rathkopf and Scher, 2013), have been used therapeutically with great benefit to prostate cancer patients; however, resistance often arises. AR PROTACs repeatedly have been shown to outperform enzalutamide, particularly in resistance contexts of increased androgen levels or mutations in AR that convert antagonists into agonists (Han et al., 2019; Salami et al., 2018). The AR PROTAC ARV-110 is currently in clinical trials for metastatic castration-resistant prostate cancer.

Estrogen Receptor

The SERDs validated induced estrogen receptor (ER) degradation as a therapeutic strategy with fulvestrant being FDA approved to treat breast cancers (Howell et al., 2004). However, ER PROTACs induce more efficient degradation than fulvestrant and possess improved pharmacological properties (Flanagan et al., 2019; Hu et al., 2019). The ER PROTAC ARV-471 is currently in clinical trials for advanced or metastatic ER+/Her2− breast cancer.

BRD4

The BET domain epigenetic reader protein BRD4 is perhaps the most commonly PROTAC-targeted protein (reviewed in (Yang et al., 2019) and is often used as a test substrate for technological advances in protein degradation (Martin et al., 2019; Spradlin et al., 2019; Ward et al., 2019; Zhang et al., 2019b). Targeting BRD4 using PROTACs is particularly attractive since its degradation results in more robust loss of c-Myc relative to simple BRD4 inhibition, resulting in enhanced apoptosis of various cancer cells both in vitro and in vivo (Lu et al., 2015; Piya et al., 2019; Winter et al., 2015). Indeed, BRD4-targeting PROTACs appear to be potential therapeutics for secondary AML (Saenz et al., 2019; Saenz et al., 2017), multiple myeloma (Zhang et al., 2018), mantle cell lymphoma (Sun et al., 2017), diffuse large B cell lymphoma (Jain et al., 2019) and castration-resistant prostate cancer (Raina et al., 2016) in preclinical models. Interestingly BRD4 PROTACs have also been employed as a test bed for PROTAC resistance mechanisms, revealing that resistance to VHL recruiting PROTACs can arise from loss of Cul2 and that CRBN recruiting PROTAC treated cells can lose CRBN or its cognate E2, UBE2G1 (Ottis et al., 2019; Zhang et al., 2019a). These mechanisms of resistance should be considered when developing PROTACs into therapeutics.

Targeting Aggregated Proteins

A challenging but potentially very exciting class of targets for protein degradation are aggregation-prone proteins such as alpha-synuclein, transthyretin and tau (Iadanza et al., 2018). Degradation of these proteins, either in their monomeric or aggregated state, via direct recruitment to E3 ligases by PROTACs represents an enticing approach to treating protein aggregation diseases. Several PROTAC studies have begun to address this challenge by inducing tau degradation (Chu et al.; Silva et al., 2019).

Biological Discoveries Enabled by PROTACs

Beyond serving as potential drugs, PROTACs provide unique tools for the study of protein function, combining the advantages of small molecules with those of other gene silencing/editing techniques. Perhaps most importantly, they enable the study of acute protein loss rather than the gradual selection of cells that survive without a given protein, thus avoiding issues arising from compensatory pathways/reprogramming. PROTACs can also be employed in contexts, such as patient samples, where other techniques may not be amenable. Below are examples where PROTACs enabled the discovery or confirmation of biological insights.

BCR-ABL independent CML Stem Cells

The development of imatinib, followed by other BCR-Abl kinase inhibitors, has transformed the lives of chronic myeloid leukemia (CML) patients. However, kinase inhibition is not curative in most patients (Druker, 2009). A population of CML stem cells survives despite the inhibition of oncogenic kinase activity (Corbin et al., 2011), and it had been postulated they depend on kinase-independent roles of BCR-Abl for survival. Unfortunately, this has proven challenging to study, especially in patient samples, due to the inherent selection requirements following genetic knockdown. Our laboratory and others have developed PROTACs capable of inducing the degradation of BCR-Abl (Lai et al., 2016; Shibata et al., 2017; Shibata et al., 2019). Recently, a collaborative study with Brian Druker employing allosteric BCR-Abl PROTACs revealed that CML patient stem cells are not dependent on the presence of BCR-Abl protein, suggesting that another mechanism drives their proliferation and alternative treatments should be explored (Burslem et al., 2019).

FLT3-ITD plays non-kinase roles in AML

Internal tandem duplication (ITD) of key FLT-3 subdomains are validated driver mutations in acute myeloid leukemia, but clinical trials have demonstrated that constant and complete kinase inhibition is required for therapeutic benefit (Grunwald and Levis, 2013; Pratz et al., 2009; Smith et al., 2012). We postulated that degradation of mutant FLT-3 ITD could provide a potential therapeutic approach since degradation results in complete and sustained lack of signaling. Conversion of the phase 3 clinical candidate quizartinib (AC220) (Zarrinkar et al., 2009) into a PROTAC resulted in a compound with enhanced antiproliferative effects relative to quizartinib despite being a less potent inhibitor in vitro and in cellulo, thus revealing non-kinase roles of FLT-3 ITD in AML (Burslem et al., 2018c).

TRIM24 as a Key Dependency in AML

Similar to BRD4 PROTAC development, PROTACs also have been used to study the bromodomain-containing transcriptional coactivating protein TRIM24. Conversion of the TRIM24 ligand, IACS9571 (Palmer et al., 2016) into a PROTAC (Gechijian et al., 2018) generated a potent TRIM24 degrader. Interestingly, while IACS9571 inhibits TRIM24 binding to hyperacetylated chromatin, it fails to induce a strong transcriptional response. However, TRIM24 degradation via PROTAC-mediated VHL recruitment significantly upregulates tumor suppressor genes in AML cells, thus demonstrating the TRIM24-dependency of acute leukemias.

BRD9 is Critical for Synovial Sarcoma

Having developed exquisitely selective BRD9 PROTACs (Remillard et al., 2017) to study this targets role in the BAF complex, the degrader was employed to confirm the dependence of Synovial Sarcoma on the non-canonical BAF complex associated with expression of the SS18-SSX oncogenic fusion protein and/or loss of SNF5 (Brien et al., 2018). Interestingly, malignant rhabdoid tumors also lack SNF-5 and rely on BRD9-containing non-canonical BAF complexes as evidenced by PROTAC-mediated BRD9 depletion, (Michel et al., 2018). This demonstrated that, upon loss of SNF-5, cancer cells remodel the BAF complex. PROTACs for other components of the BAF complex have also recently been reported (Farnaby et al., 2019).

FAK Scaffolding Roles are Crucial for Cell Migration but Not Proliferation

Given that patient studies with Focal Adhesion Kinase (FAK) inhibitors have failed to live up to their preclinical promise, Cromm et al. incorporated the FAK inhibitor, defactinib, into a PROTAC to assess the effect of FAK degradation versus inhibition (Cromm et al., 2018). Cell culture studies revealed that, while FAK kinase inhibition is unable to prevent cell migration in wound healing assays and invasion in trans-well assays, FAK degradation was efficacious thereby underscoring the kinase-independent functions of FAK. Interestingly, both this work and a concurrent study from Boehringer Ingelheim (Popow et al., 2019) failed to phenocopy the antiproliferative effect of shRNA-mediated FAK depletion (McDonald et al., 2017).

Tag-Based Systems

The development of a bespoke PROTAC for a potential target may be beyond the means of many academic laboratories, therefore tag-based methods have been developed to combine genetic modification with the power of the PROTAC technology. The two most commonly used PROTAC systems are discussed below.

HaloPROTACs

The HaloPROTAC system utilizes the HaloTag® protein, an engineered bacterial dehalogenase, which allows the orthogonal, covalent conjugation of a chloroalkane-labelled molecule to a fusion protein of interest (Los et al., 2008). Both VHL- (Buckley et al., 2015; Tovell et al., 2019) and cIAP-recruiting HaloPROTACs (Tomoshige et al., 2016; Tomoshige et al., 2015) have been reported to induce degradation of various HaloTag fusion substrates, including cytosolic (ERK, MEK, GFP), endosomal (VPS34, SGK3) and nuclear-localized (CREB1) proteins. The HaloPROTAC system has also afforded biological insights into multiple systems, as detailed below.

The Role of PNPLA3 in Fatty Liver Disease

BasuRay et al. used HaloPROTAC3 to degrade HaloTag fused with patatin-like phospholipase domain-containing protein 3 (PNPLA3) in vivo (BasuRay et al., 2019). An I148M mutation in PNPLA3 is associated with non-alcoholic fatty liver disease, which leads to steatosis via the accumulation of triglycerides into lipid droplets (Smagris et al., 2015). The I148M mutation results in an approximately 80% reduction in triglyceride hydrolase activity, but surprisingly the presence of the reduced or non-catalytic PNPLA3 protein is yet required for the development of the hepatic steatosis. PROTAC-mediated in vivo degradation of I148M PNPLA3 returned liver triglycerides in mouse to normal, providing additional evidence that mutant PNPLA3 accumulation is responsible for hepatic steatosis.

Kinetics of WASH Complex Formation

The HaloPROTAC system has also been used to confirm that heat shock factor binding protein 1 (HSBP1) is an assembly factor for the WASH complex, which plays a key role in endosomal sorting. Degradation of a HaloTag:WASH fusion protein and subsequent siRNA knockdown of HSBP1 during PROTAC washout demonstrated that HSBP1 is the assembly factor required for remodeling of CCDC53 homotrimers into the WASH complex (Visweshwaran et al., 2018). This approach highlights the use of PROTACs to enable temporal control of protein levels, thus enabling pulse chase experimentation that may otherwise be challenging.

dTAG

The dTAG system works in similar way to HaloPROTACs but instead of tagging protein with the HaloTag protein, the F36V FKBP mutant protein (Clackson et al., 1998) is fused to the protein of interest. The dTAG system uses an F36V-selective “bump” ligand, tethered to a IMiD derivative to recruit CRBN and subsequently degrade the FKBP-fusion protein (Nabet et al., 2018). This system has been shown amenable to in vivo experimentation via IP injection at 25 mg/kg. Proof-of-principle studies have shown dTAG amenable to a wide variety of proteins including HDAC1, MYC, EZH2, PLK1 and KRAS^G12V and the system has been employed to address biological questions as exemplified below.

Basal-like Breast Cancer Cells are not MELK Dependent

Previous studies using shRNA silencing suggested that Basal-like Breast Cancer (BBC) cells depend on MELK expression for survival (Hebbard et al., 2010; Touré et al., 2016). However, a thorough study employing selective MELK inhibitors, CRISPR and PROTAC-mediated degradation discovered BBC cells are agnostic to MELK levels (Huang et al., 2017). Huang et. al. employed CRISPR gene editing to knockout MELK and observed no effect on proliferation. Similarly, MELK inhibitors exhibited no antiproliferative activity. Concerned that compensatory signaling arose during the selection/cloning procedures to explain their observations, they also introduced a dTAG version of MELK prior to knockout of endogenous MELK such that MELK was constantly expressed circumventing any impetus for compensation to arise. Employment of a dTAG PROTAC to rapidly and selectively induce dTAG-MELK degradation showed no effect on proliferation despite acute MELK loss, confirming that BBC cells are not MELK-dependent.

Cytosolic Mutant Nucleophosmin is Crucial for Leukemogenesis

Goodel and co-workers used CRISPR/Cas9 gene editing to demonstrate that mutant nucleophosmin (NPM1) relocalization, via nuclear export sequence disruption, could suppress cell proliferation and induce differentiation of hematopoietic stem cells (Brunetti et al., 2018). This occurs via disruption of the HOX/MEIS1 transcriptional program, consistent with the hypothesis that HOX/MEIS1 genes are master regulators of the hematopoietic lineage (Argiropoulos et al., 2007). To further confirm that lack of cytosolic NPM1 induces this phenotype, Brunetti, Gundry et. al. employed the dTAG system to deplete NPM1 rapidly (>85% loss in 4 hours). This resulted in the same phenotype, thereby confirming direct correlation between HOX/MEIS1 expression downregulation and lack of cytosolic NPM1.

The dTAG system has also been used to confirm the addiction of acute myeloid leukemia cells to ENL (YEATS domain-containing protein 1) following its identification in a genome wide CRISPR screen (Erb et al., 2017) and to demonstrate that degradation of SNF5 allows the re-association of MYC with chromatin (Weissmiller et al., 2019). It has also been used to demonstrate that YY1 has no direct role in the modulation of replication timing (Sima et al., 2019) and to demonstrate that the OCT4 is crucial for the localization of Mediator condensates at super-enhancers in embryonic stem cells (Boija et al., 2018).

HaloPROTAC vs dTAG

Both HaloPROTAC and dTAG are powerful approaches and could conceivably be used simultaneously due to their orthogonality. However, there are subtle differences between the two systems which may determine which system is most suitable for each application. For example, dTAG requires a smaller tag (FKBP12^F36V - 12 kDa) be incorporated into the target protein than Halotag (33 kDa), which may favor the use of dTAG in congested systems where tags may perturb protein-protein interactions. However, the commercially availability and multitude of other uses for Halotag fusions (England et al., 2015) makes it the perfect choice if, for example, one wishes to look at subcellular localization of a protein as well as induce its degradation. Furthermore, the HaloPROTAC system does not exhibit the ‘hook effect’ observed with dTAG. Additionally, the use of VHL ligands in the HaloPROTAC approach enables the use of diastereomer controls and avoids issues associated with the use of IMiD based PROTACs discussed below.

Workflow

In order to use the PROTAC system, it is necessary to either develop de novo a PROTAC capable of targeting the protein of interest, or to modify the latter with a tag (described above) to enable degradation via HaloPROTAC or dTAG degrader molecule. Indeed, it may be advantageous to perform preliminary studies with the tagged target protein to establish proof-of-concept prior to developing a PROTAC to target the endogenous protein. Given the recent advances in genome editing, it is relatively easy to tag a protein of interest at an endogenous locus to enable its ligand-induced degradation without the necessities of target overexpression or developing novel ligands (Brand and Winter, 2019). Displayed in Figure 3 is a typical workflow to develop a PROTAC. The workflow begins, obviously, with selecting a target protein to degrade.

Figure 3 –

PROTAC Development Workflow

Target Selection

Some ideal PROTAC targets are 1) proteins without enzymatic function and thus cannot be modulated with traditional small molecules, 2) proteins that have additional roles beyond enzymatic activity (e.g. scaffolding roles of FAK or BCR-ABL), or 3) proteins that, upon inhibition, undergo compensatory upregulation, thus making it difficult to achieve maximal loss of protein function. It may be useful to look at previous knockdown experiments (e.g., phenotypic genome-wide CRISPR screens or RNAi) or for cell type-specific dependencies during target selection to gather information about the target protein (Tsherniak et al., 2017).

There are some proteins for which inhibition is sufficient to abrogate all function and degradation adds little in terms of biology but these proteins are likely few and far between. In these cases, PROTACs can nevertheless be beneficial in terms of reduced dosage and duration of effect. More excitingly, PROTACs can extend the druggable proteome to any protein which is ligandable. Of the approximately 17470 observed proteins (Omenn et al., 2018), only 10–15% are considered druggable (Hopkins and Groom, 2002) but many more have been shown to be ligandable (Backus et al., 2016; Parker et al., 2017) yet many of these small molecule binding events enact no function. The ability to impart activity to otherwise biologically inert ligands via PROTAC conversion greatly expands the druggable proteome.

PROTAC Development

Subsequently, a thorough literature investigation for known target ligands should be performed. If a ligand is available, inspection of either a co-crystal structure or known structure-activity relationships may reveal a suitable location for linker attachment. At this point PROTAC conversion can begin, employing synthetic/medicinal chemistry to append a linker and E3 recruiting element to the targeting ligand. If no ligand has previously been reported, or the reported compounds have questionable structures, then a campaign to identify a ligand for the target protein may be initiated; alternatively, it may be simpler to use the HaloPROTAC or dTAG approach initially. We recommend the use of multiple linker lengths and compositions when developing PROTACs, especially if the synthetic approach allows for modularity. Linker length and composition can have profound effects, for example HaloPROTACs with an ethylene glycol unit removed have no ability to induce degradation (Buckley et al., 2015), BCR-ABL PROTACs with an ether replacing an amide demonstrate more cell permeability (Burslem et al., 2019) and Lapatinib-based PROTACs can be either dual EGFR/Her2 degraders or selective EGFR degraders depending on linker length (Burslem et al., 2018b).

Once the PROTAC candidate molecules or cell line expressing tagged proteins are available, it is crucial to confirm firstly that they do indeed induce the target degradation. Most commonly this is achieved by immunoblotting, although mass spectrometry or flow cytometry can be used depending on the target protein properties (Buckley et al., 2015). Iterative rounds of screening may be necessary to synthesize a sufficiently potent PROTAC or it may be necessary to explore several tagging approaches/sites of tag incorporation into the target for maximal degradation. We generally recommend an initial PROTAC treatment time of 24 hours, but kinetic characterization often reveals degradation occurs much more rapidly. Recently, bespoke technologies for the study of PROTACs in cells have been developed that may prove instrumental in guiding their development (Riching et al., 2018). The synthesis and assay of candidates is normally the rate limiting step for the use of a PROTAC and depends on the quality of the chemical matter available for the target. Starting from a well characterized small molecule provides a distinct advantage and as more become available via the Target 2035 initiative (Carter et al., 2019), the rate and ease of PROTAC development will no doubt increase.

PROTAC Validation

Once a PROTAC is identified, it is crucial to carefully characterize the degradation event via control experiments -- e.g., via qPCR confirmation that protein loss occurs at a post-translational level and not by decreases in mRNA (Bondeson et al., 2015) -- prior to its use as a tool. Some small molecules induce apparent degradation of their target protein, but are actually acting at the transcriptional level (Field et al., 2017). It is also crucial to confirm that degradation is occurring via the UPS. To do this, cells can be co-treated with pharmacological modulators of the UPS. We specifically recommend co-treatment with a proteasome inhibitor (epoxomicin (Kim and Crews, 2013) or bortezomib (Paramore and Frantz, 2003)) which should preserve protein levels at untreated levels. Additionally, co-treatment with MLN-4924 (Pevonedistat), a NEDD8-activating enzyme inhibitor, will confirm that an active E3 ligase is involved in the degradation (Cullin ligases require neddylation for activity) (Brownell et al., 2010). Due to the toxicity of these control compounds, it is often necessary to use a shortest incubation time required for significant degradation (e.g. 5 hours vs 24 hours).

Importantly, degradation by application of a small molecule enables the use of an inactive analogue as a control compound. This is crucial for the discovery of novel biology when employing a PROTAC derived from an inhibitor. Typically, an inactive PROTAC can be created by discrete modification of the E3 ligase ligand structure to compromise its activity: the resulting molecule is incapable of inducing degradation, yet equally potent as the PROTAC at inhibiting the target. We favor the use of the VHL-recruiting ligand for careful interrogation of inhibition versus degradation since the simple inversion of stereocenters on the VHL ligand yields a compound with equal inhibitory activity and pharmacological properties, such as cell permeability, thus enabling it to function as a true control (Burslem et al., 2018b). Inversion of stereocenters can also be employed to generate inactive compounds as controls for MDM2-recruiting PROTACs (Hines et al., 2019). It is possible to compromise the CRBN-recruiting IMiD ligands via omission of a carbonyl or by methylation of the imide functionality, however this does subtly modulate its physicochemical properties (Burslem et al., 2018a; Huang et al., 2018). Care must be taken when using IMiD analogues as they may still induce degradation of zinc-finger neo-substrates associated with IMiD pathophysiology (Ishoey et al., 2018). Inactive PROTACs are important experimental controls that can confirm/disprove the PROTAC mechanism of degradation given that ligand binding alone could destabilize the target without necessarily recruiting an E3 ligase (Huang et al., 2017), such as is the case with hydrophobic tagging or SERDs (Gustafson et al., 2015; Neklesa et al., 2011; Wardell et al., 2011).

An additional, optional experiment to conduct at this point is a proteomic characterization of the effect induced by PROTAC treatment. Mass spectrometry or reverse phase protein arrays for example, can enable the user to acquire selectivity data for the PROTAC. Quantitative proteomics is a powerful tool to identify other proteins, beyond that targeted, which are downregulated by PROTAC treatment. This may, of course, be a biological result of loss of the target protein such as with c-Myc loss when BRD4 is degraded (Lu et al., 2015; Winter et al., 2015) or a pharmacological result such as the unexpected degradation of p38α by foretinib-based PROTACs (Bondeson et al., 2018; Smith et al., 2019).

Advantages and Disadvantages of PROTACs

Advantages

Catalytic Mechanism of Action

Since PROTACs operate via an event-driven rather than occupancy-driven mechanism, they act as catalysts for selective protein degradation in that one molecule of PROTAC induces the ubiquitination of multiple molecules of target protein (Bondeson et al., 2015). Indeed, recent studies demonstrate that as little as 10% E3 ligase occupancy is able to efficiently induce target degradation (Zhang et al., 2019b), albeit with a covalent modification of the E3 creating a permanently reprogrammed E3 ligase. Limited occupancy on the targeting ligand side has also been demonstrated to be sufficient via the use of low affinity ligands (Crew et al., 2018; Smith et al., 2019) or co-treatment with a competitive agonist (Salami et al., 2018). This catalytic nature of PROTACs mimics RNA interference.

Small Molecule Nature

Their small molecule nature makes PROTACs as simple to use in experiments as inhibitors. No specialized transfection reagents, culture conditions or viruses are required for PROTAC application. Indeed, with the exception of the tag-based systems, the use of PROTACs requires no genetic modification of the model system allowing the interrogation of completely endogenous systems. This allows for the direct comparison of degradation versus inhibition with the same methodology, resulting in an enhanced biological understanding of protein function and abrogating the risks of perturbing complex biological systems with, for example, transfection reagents that may either induce effects of their own or mask subtle effects RNAi(Jacobsen et al., 2009).

Another advantage arising from the small molecule nature of PROTACs is control over the concentration of compound enabling the quantitative control of protein levels. Rather than employing a “digital” on/off switch such as is observed with gene editing, a PROTAC can be used in an “analogue” fashion to modulate protein levels between 0–100%. This may be particularly powerful to address proteins that are overexpressed in a disease state, but nevertheless essential in a normal/healthy state. Following a dose-response experiment, it is possible to treat at a dose that induces sub-maximal degradation, potentially restoring overexpressed protein levels down to normal.

Temporal Control

PROTACs allow exquisite temporal control of protein levels - degradation can be observed in as little as 1 hour allowing the study of acute protein loss in a way not possible with many other techniques. This prevents biological compensation from arising upon deletion of a key protein during selection as happens with various RNA interference techniques as well as CRISPR/Cas9. A common CRISPR technique is to employ Homology Directed Repair (HDR) to insert GFP or resistance genes at the perturbed loci (Leonetti et al., 2016), allowing the selection of cells that were successfully edited. While undoubtedly powerful, this multi-day selection could conceivably select a cellular population with an alternative pathway driving proliferation or circumnavigation of issues arising from loss of a particular protein. PROTACs are able to rapidly deplete a protein in an entire cellular population providing a more accurate method to interrogate functions of a protein in a native context.

Additionally, PROTAC withdrawal allows for the rapid restoration of target protein levels as fast as protein re-synthesis permits. Thus, this provides a reversibility enabling elegant experiments such as those described above for the WASH complex. Although some PROTAC may survive inside cells after wash out of compound and continue to enact degradation, concurrent addition of excess E3 ligase ligand can competitively prevent any additional degradation from occurring (Burslem et al., 2018b).

Portability

A further benefit of the PROTAC approach lies in its portability. Generally, if a PROTAC induces ubiquitination and degradation of a protein within one system, it can be applied more broadly providing the new system has the required machinery (E3 ligase etc.). This enables the rapid screening of protein roles in different cell types without requiring their genetic modification. Furthermore, it enables the study of proteins in contexts not amenable to other techniques, such as unculturable patient cells, e.g., CML stem cells discussed above (Burslem et al., 2019).

Another advantage of portability lies in the transition to in vivo experiments and potentially clinical translation. Application of PROTACs in vivo requires no genetic manipulation of the animals, enabling both more rapid progression of experimentation and the ability to probe unperturbed systems. While PROTACs may need optimization of their physico-chemical properties for in vivo work, there are many examples of PROTACs functioning in vivo, including mammals (Bondeson et al., 2015; Burslem et al., 2018c; Jain et al., 2019; Nabet et al., 2018; Winter et al., 2015) and non-human primates (Sun et al., 2019). Despite any potential pharmacokinetic challenges with PROTACs, their small molecule nature and portability make them readily and directly translational in a way that other techniques are not. It is important to note that PROTACs often exhibit better pharmacokinetic properties than would be anticipated from their molecular weight. In fact, is has been possible to develop PROTACs with oral bioavailability in humans as evidenced by the recent reports from the Phase 1 clinical trials of ARV-110 and ARV-471 which achieved exposures in the efficacious range observed in preclinical studies when administered orally once per day.

Disadvantages

Discovery Phase

Despite all the advantages outlined above for PROTACs, they are not without their associated challenges. One major disadvantage is the lead time for PROTAC development, which can be relatively lengthy compared to designing and ordering an siRNA or gRNA. As outlined in Figure 3, PROTACs require not only a ligand for the protein of interest but also its conversion into a PROTAC. This can be a time-consuming process involving substantial synthetic/medicinal chemistry. The tag-based systems (HaloPROTACs and dTAG) bypass this discovery phase, but also abrogate some of the advantages of the PROTAC system concerning portability and lack of genetic manipulation.

Off-Target Effects

As with other techniques, off-target effects with PROTACs are possible. Following PROTAC treatment, the use of proteomics to quantify proteins, including those expressed at low levels, provides a powerful tool to assess the off-target effects of PROTACs (Savitski et al., 2018). Even if the recruiting element selectivity is known, proteomics can reveal surprising, new PROTAC substrates likely resulting from the additive effects of protein-protein interactions between the protein in question and the E3 ligase (Bondeson et al., 2018). In theory, ligand binding to an E3 ligase could perturb binding of endogenous substrates. Fortuitously, much higher ligand concentrations than are typically employed with VHL-recruiting PROTACs (Frost et al., 2016) are required to stabilize its endogenous target, HIF-1α (Burslem et al., 2017); however, off-target effects could be more problematic with other E3 ligases for which the native substrates currently are poorly-characterized or unknown. The off-target effects of CRBN-recruiting PROTACs must be very carefully assessed, since the IMiD components, alone or incorporated into PROTACs, can induce the degradation of zinc finger CRBN neo-substrates (Fischer et al., 2014; Ishoey et al., 2018; Krönke et al., 2014; Matyskiela et al., 2016).

Hook Effect

PROTACs, and other bifunctional molecules, exhibit an initially curious but entirely rational phenomenon whereby higher concentrations do not always result in more effect (Douglass et al., 2013). This so called “hook effect” results from the formation of unproductive dimers at high PROTAC concentration rather than the productive trimeric complex required for degradation. This leads to concerns surrounding PK/PD relationships and dosing regimens. However, it should be noted that favorable protein-protein interactions between the E3 and target protein appear to increase the maximum concentration that can be used before the hook effect is observed (Buckley et al., 2015; Tovell et al., 2019).

Not all proteins/subcellular locations amenable yet

Finally, since PROTACs are still an emerging technology, they are not yet demonstrated to function against every protein class nor in every subcellular compartment (Figure 4). Cytosolic and nuclear proteins can routinely be degraded (Bondeson et al., 2015); indeed some E3 ligases allow selective nuclear degradation (Zhang et al., 2019b). Several examples of receptor tyrosine kinase degradation have been reported (Burslem et al., 2018b; Burslem et al., 2018c), but no examples of more than single pass transmembrane proteins have yet been reported. HaloPROTACs have been employed to degrade proteins localized to the membranes of endosomes (Tovell et al., 2019). Despite hydrophobic tagging experiments having validated unfolded protein responses in both the Golgi and the ER (Hellerschmied et al.; Raina et al., 2014; Serebrenik et al., 2018), degradation of proteins via PROTAC has yet to be reported therein. Bespoke ligands for ligases localized to these organelles will likely be required to enact PROTAC-mediated degradation in these subcellular locations (Smith et al., 2011).

Figure 4 -.

Diagram of a typical mammalian cell denoting locations of proteins which are susceptible to PROTAC mediated degradation. Solid lines represent locations which have been targeted with PROTACs. Dashed lines represent locations which have not yet been targeted with PROTACs.

Comparison to Alternatives

In this section we will briefly compare PROTACs with other current technologies that enable the study of protein function and discuss applications where one technique may be preferable.

RNA Interference

The use of RNA interference has become ubiquitous in modern biological research and represents an excellent tool to study protein function (Agrawal et al., 2003; Setten et al., 2019). PROTACs in many ways function as chemical siRNA equivalents, although they do differ slightly. PROTACs have the advantage of being experimentally less complex than using siRNA since no transfection reagents are required. However, the ease of PROTAC use is juxtaposed with their rigor of development. Thus, for large target libraries, siRNA may be advantageous whereas for the study of one particular protein, PROTACs have the advantage. Both techniques represent “knockdown” approaches and suffer from potential off-target effects but given the common use of proteomics to characterize PROTACs the off-targets may be better known. Indeed, PROTACs have been used to invalidate hits from siRNA experiments (Huang et al., 2017; Nunes et al., 2019; Popow et al., 2019). It may also be possible to achieve greater overall reduction in protein levels, particularly for long-lived proteins, with PROTACs compared to RNAi since PROTAC activity target the protein rather than preventing additional expression.

Gene Editing

CRISPR/Cas9 and related gene editing techniques are certainly efficient in experiments where long-term protein depletion is required. However, the requisite selection is both time consuming and allows for biological compensation to the protein loss. PROTACs provide an advantage in terms of kinetics, allowing the study of acute protein loss to expedite workflow and potentially provide more relevant details about the endogenous system. Gene editing provides protein knockout as long as each copy of the gene is edited, this is challenging in polyploid cells (e.g. HeLa), but since PROTACs act post-transcriptionally this is not a concern with their use.

Post-Translational Protein Degradation

There are many other methods for small molecule induced post-translational protein degradation (for example auxin-induced degradation and the Shield-1 system), which we have reviewed in detail elsewhere (Burslem and Crews, 2017). However, these rely on the expression of tagged proteins and/or expression of other endogenous proteins; PROTACs do not. Recently, the Trim-Away system was reported (Clift et al., 2017) employing antibodies to target the E3 ligase TRIM21 to proteins, leading to their degradation. However, this requires the expression of TRIM21 protein and for the antibody to be microinjected or electroporated into the cell, thus lacking the small molecule advantages of PROTACs.

Conclusion

In this Primer we demonstrate, with examples and an explanation of the technology, the power of PROTACs as tools to probe biological systems as well as to be potential therapeutics. This targeted protein degradation approach adds a new tool in the experimental biology toolbox, combining the benefits of RNAi and small molecule inhibitors and provides complementary orthogonality to the preexisting tools. We hope that PROTACs will become a mainstay in protein function investigation and we believe they enable us, as a community, to address biological questions which are currently intractable. For example, the reversibility and kinetic advantages of PROTACs provide tools to temporally control protein levels with much higher resolution that other approaches. This provides the opportunity to compare acute versus chronic loss of a protein and to study the effect of reintroducing that protein. The ability to deplete a protein for a defined time and then restore it to endogenous levels as fast as transcription allows may prove useful in the study of protein complex assembly, particularly with respect to order of association. As the PROTAC technology continues to mature the disadvantages described above will become less significant. For example, the current largest challenge in the use of PROTACs is the length of the discovery phase. As we continue to develop enhanced knowledge of the UPS, of the requirements for potent PROTACs and of ligands for protein targets, the discovery phase will be significantly shorter. Similarly, as ligands for additional E3 ligases become available, the likelihood of successful PROTAC development increases and more cellular locations become available.