Prey for the Proteasome: Targeted Protein Degradation—A Medicinal Chemist's Perspective

Abstract Targeted protein degradation (TPD), the ability to control a proteins fate by triggering its degradation in a highly selective and effective manner, has created tremendous excitement in chemical biology and drug discovery within the past decades. The TPD field is spearheaded by small molecule induced protein degradation with molecular glues and proteolysis targeting chimeras (PROTACs) paving the way to expand the druggable space and to create a new paradigm in drug discovery. However, besides the therapeutic angle of TPD a plethora of novel techniques to modulate and control protein levels have been developed. This enables chemical biologists to better understand protein function and to discover and verify new therapeutic targets. This Review gives a comprehensive overview of chemical biology techniques inducing TPD. It explains the strengths and weaknesses of these methods in the context of drug discovery and discusses their future potential from a medicinal chemist's perspective.


Introduction
Classical approaches in the drug discovery process typically aim to identify high affinity small molecules modulating the activity of target proteins.T his happens in an occupancy-driven manner with the inhibitor binding to its target and occupying the binding site,t hereby inhibiting target function. [1] This strategy has been very successful for many targets harboring atractable active or allosteric site like enzymes or receptors.However,the majority of proteins still remains challenging to address,r endering 80-85 %o ft he human proteome undruggable due to the absence of available binding pockets and suitable chemical matter. [2] Targeted protein degradation (TPD) offers an ovel therapeutic alternative by inducing the depletion or reduction of ad iseasecausing protein via hijacking the endogenous protein degradation machineries.T PD has the potential to target the undruggable proteome that limits current drug discovery efforts,a so nly ab inder is required to recruit the target protein for degradation rather than high affinity inhibitors. [3] In addition to their therapeutic potential, protein degraders are valuable chemical biology tools to validate targets and to gain ad eeper understanding of protein function and cellular pathways.A na dvantage of targeted protein degradation is the acute and rapid depletion of target proteins avoiding unwanted adaption events in comparison to RNAorgenome editing tools and the ability to deplete proteins with as low turnover rate.I nc ontrast to nucleotide-based methods like RNAi, antisense oligonucleotides or genome editing strategies which all suffer from limited therapeutic applications due to their low in vivo stability and bioavailability, [4] TPD harnesses more drug-like small molecules.T hese degrader molecules are typically inducing novel protein-protein interactions (PPIs) and can either comprise hetero-bifunctional molecules such as proteolysis-targeting chimeras (PRO-TACs) [3,5] or monovalent molecular glues [6] which are of non-chimeric nature.T his unique mode of action combining target engagement with subsequent degradation allows them to address targets previously out of reach.
Native protein degradation occurs via one of two main mechanisms in the cell:t he ubiquitin-proteasome system (UPS) and the autophagy-lysosome pathway ( Figure 1). Proteins are marked for degradation by ac ovalent post-translational modification with the 76 amino acid protein ubiquitin. [7] Covalent attachment of ubiquitin to the substrate follows at hree-step mechanism involving E1, E2 and E3 enzymes (Figure 1a). [8] Subsequent sequential ubiquitination of the conjugated ubiquitin results in the formation of apolyubiquitin chain. Ubiquitin contains seven lysine (Lys) residues for polyubiquitin chain formation (Lys 6, Lys11, Lys27, Lys29, Lys33, Lys48, or Lys63). While Lys48and Lys11linkages mediate proteasomal degradation, Lys63-linked ubiquitin chains are prone to degradation via autophagy. [9] While the 26S proteasome efficiently degrades shortlived, soluble unfolded or misfolded proteins and polypeptides, [10] the autophagy-lysosome system is responsible for elimination of long-lived proteins,i nsoluble protein aggregates and dysfunctional organelles such as degenerated mitochondria. [11] There are three main forms of autophagy: macroautophagy,m icroautophagy and chaperone-mediated autophagy. [12] In this review,t he focus is on macroautophagy (here referred to as autophagy). Autophagy is characterized by the formation of ad ouble-membrane structure,t ermed autophagosome,w hich sequesters cytosolicp roteins,p rotein aggregates or dysfunctional organelles (Figure 1b). [13] These autophagosomes subsequently fuse either with endosomes or directly with lysosomes to form autolysosomes,u ltimately resulting in the destruction of their contents by hydrolytic enzymes. [12] Targeted protein degradation (TPD), the ability to control aproteins fate by triggering its degradation in ahighly selective and effective manner,h as created tremendous excitement in chemical biology and drug discoverywithin the past decades.The TPD field is spearheaded by small molecule induced protein degradation with molecular glues and proteolysis targeting chimeras (PROTACs) paving the way to expand the druggable space and to create an ew paradigm in drug discovery.However,besides the therapeutic angle of TPD aplethora of novel techniques to modulate and control protein levels have been developed. This enables chemical biologists to better understand protein function and to discovera nd verify new therapeutic targets. This Review gives acomprehensive overview of chemical biology techniques inducing TPD.Itexplains the strengths and weaknesses of these methods in the context of drug discoverya nd discusses their future potential from amedicinal chemistsp erspective.

Conclusion and Outlook 15461
Ubiquitination is at the heart of the degradation machinery for proteasomal-as well as autophagy-mediated degradation, with E3 ligases as the critical components of the ubiquitination cascade.E 3l igases provide strict spatial and temporal substrate specificity and over 600 genes encoding for ubiquitin E3 ligases in the human genome,contributing to the cellular specificity and complexity of the ubiquitination system. [14] E3s have been structurally classified in really interesting new gene (RING), Cullin-RING-, U-Box-and homologous to the E6-AP carboxyl terminus (HECT)-type E3 ligases and each class employs ad ifferent mechanism for ubiquitin conjugation. [15] Thel argest superfamily of E3 ligases,t he Cullin-RING E3 ligases,a re multi-subunit modular complexes which simultaneously bind the target protein and the E2 enzyme,e nabling the transfer of ubiquitin from the E2 protein to the target. They consist of four main components,aCullin protein serving as as caffold protein, aR ING finger protein which can bind to the E2 enzyme, as ubstrate receptor recognizing the target and an adaptor protein bridging the receptor with the Cullin protein (Figure 1c). HECT E3 ligases receive the ubiquitin on an active site cysteine residue from the E2 enzyme before transferring it to their target proteins.O nly about 1% of all E3 ligases have been explored in TPD to date with Cullin-RING E3 Ligases being the main E3s exploited so far. Predominantly, the E3 ligase receptor proteins Cereblon (CRBN) and von Hippel Lindau (VHL) are utilized for TPD followed by the inhibitor of apoptosis protein (IAP) and the E3 ligase mouse double minute 2( MDM2). Increasing the knowledge of structural properties and expression pattern as well as the repertoire of E3 ligase binders bears the potential to expand the application of E3 ligases in therapeutic drug discovery, particularly TPD.
This review provides acomprehensive overview of various TPD techniques for chemical biology and drug discovery, highlighting their strengths and limitations from am edicinal chemistsp erspective.T od iscriminate between the different methods,w ed ivided them into two main categories:1 .Degradation of tagged target proteins;2 .Degradation of untagged target proteins.Aprotein tag comprises ap eptide sequence or protein genetically fused to the protein of interest (POI) to allow its selective detection and/or purification. [16] In doing so,t he tag allows highly selective recognition of the POI without the need to identify asuitable binder like as mall molecule,p eptide or antibody.W hile no suitable target protein binder needs to be identified using tagged proteins comes at the cost of modifying the genome of the target cell/organism in advance,l imiting the degradation of tagged target proteins predominantly to the field of chemical biology and target validation. In contrast, the degradation of untagged proteins allows studying the target in its native state but is dependent on the identification of as uitable target binder.H owever,a si ti si ndependent of genetic alterations to the organism, it is universally applicable.A nalog to ad rug discovery process the introduction of tagged TPD tools are evaluated first (section 2), which allow target validation and target discovery.I nt he following the focus is on untagged TPD strategies (section 3), that are based on specific binders and hold great potential for future therapeutic applications.
Laura Luh obtained her PhD working on transcription factors with implications in cancer at the Goethe University Frankfurt. After joining Evotec in Hamburg to establish new techniques to assess small molecule target interactions, she moved to Cambridge to join Anne Bertolotti's lab at the MRC Laboratory of Molecular Biology conducting her postdoctorals tudies on specific phosphatase inhibitors in order to correct protein misfolding defects. In 2018, she joined Bayer to work on targeted protein degradation in different therapeutic areas.
Ulrike Scheib studied biochemistry in Tübingen and completed her master's thesis at the UC in San Diego. Forher PhD in biophysics she joined the optogenetic research group of Peter Hegemann at the Humboldt-University of Berlin in 2013 and studied rhodopsin-guanylyl cyclase. In 2018 she started as aPostDoc at Bayer in Berlin and is particularly interested in structurefunction relationships and protein degradation.
Katrin Juenemanns tudied biology at the Friedrich-Schiller-UniversityinJ ena. She completed her PhD in Molecular biology at the Leibniz Institute on Aging in Jena and continued her scientific work on neurodegenerative diseases as apostdoctoral fellow at the Academic Medical Center in Amsterdam. In 2016 she joined the Leibniz Institute for Molecular Pharmacology as an AXA Research fellow to study intracellular ubiquitination and protein degradation associated with neurodegeneration. As ascientist in preclinical research at Bayer she works on targeted protein degradation.

Degradation of Tagged Target Proteins
To generate ab etter target understanding and validate at arget in ad isease context controlling protein activity and abundance is av ery powerful tool. Although, methods like CRISPR/Cas [17] have dramatically impacted the target identification and validation workflow,i nducing the degradation of the target protein in afast, controlled and selective manner provides additional benefits. [5d, 18] At the target discovery and validation stage of adrug discovery program there is typically no suitable binder available which allows to aim for degradation of the endogenous protein. This problem can be eluded by fusing aprotein tag to the target protein enabling selective recognition of the tagged protein. Avariety of tags,r anging from short amino acid sequences to protein domains are available to be introduced at the N-or C-terminus of the POI. [16] While some tags solely allow selective addressing of the target protein, others are directly inducing its degradation. [19] In order to minimally perturb the system in question the protein tag should be endogenously inserted into the host genome using for example,C RISPR/Cas.A lternative methods,s uch as transfection/transduction of aD NA-encoding plasmid in an either transient or stable fashion, might be considered as well but must be handled with care as they produce overexpressed protein levels which represent deviations from the native system. Additionally,i nc ases where protein overexpression is used the endogenous native protein levels are still present in the cell and can provide ac onstant basal activity.

Degrons
Maintaining protein homeostasis is pivotal for cell survival and proliferation. Therefore,t he equilibrium between protein synthesis and degradation must be carefully regulated to keep the system in balance.P roteins,w hich are naturally degraded by the UPS,a re displaying specific degradation signals,t ermed degrons,w hich represent E3 ligase binding motifs. [10b,20] Degrons can range from as ingle amino acid to short peptide sequences or domains and are transferable between proteins.H ence,a ttaching ad egron to any other protein induces its degradation. Several constitutive and conditional degrons,w hich can be activated by small molecules,light or temperature,have successfully been utilized to induce target degradation. [19, 20b] 2.1.1. Temperature Sensitive Degrons One of the first conditional degrons was identified in budding yeast and allows to induce temperature dependent protein degradation (Figure 2a). This is achieved by fusing at emperature sensitive dihydrofolate reductase (ts-DHFR, 25 kDa) to the target protein. Thets-DHFR-POI construct is stable at room temperature (23 8 8C-25 8 8C) but starts to unfold at physiological temperature (37 8 8C) inducing protein degradation. [21] While this temperature-mediated POI degradation is ap owerful tool for budding yeast its use in other species is limited. Atemperature difference affects the whole cell/organism and can lead to undesirable off-target effects making them challenging to apply in higher order organisms.

Auxin-Inducible Degron
Aw idely used and robust method to degrade ab road range of target proteins is the auxin-inducible degron (AID). In plants,t he hormone auxin triggers the degradation of the AID by means of the plant F-box protein TIR1 and the CUL1-RING ligase. [22] Auxins,i ndole-3-acetic acid or 1naphthaleneacetic acid, bind to TIR1 in amolecular glue like fashion and recruit AIDs for degradation ( Figure 2b). [23] As most proteins of the UPS are conserved in eukaryotes only TIR1 as well as the AID need to be ectopically introduced. Consequently,t he AID system is bio-orthogonal as neither TIR1 nor auxin are naturally present in mammalian cells. [5d] However,insome cases high concentrations of tryptophan or related metabolites can lead to some basal target degradation even in the absence of exogenous auxin. Thus far, the AID system has successfully been used to rapidly degrade both nuclear and cytoplasmic AID-fused target proteins in various species as well as in human cell lines. [22,24] In order to reduce unwanted auxin-independent TPD and to reduce the size of the AID (25 kDa) am uch leaner version termed mini-AID (7.4 kDa) was developed which can be efficiently introduced into the host genome using CRISPR/Cas. [25] Despite the successful use in mammalian cells and the development of mini-AID,the AID system is not currently amendable to use in vivo in mammals as auxin has to be used at very high concentrations.C onsequently,t his makes the design of ahigher-affinity AID very desirable.

Destabilizing Domains
As mentioned above the AID demands,inaddition to the AID itself,t he exogenous introduction of the F-box protein TIR1 to achieve degradation. An alternative presents the use of destabilizing domains (DDs) which constitutively induce the degradation of the target protein themselves.Upon ligand binding the DD is stabilized in adose dependent manner thus preventing the degradation of the DD-POI fusion construct ( Figure 2c). Thus far,D Ds have been described for several protein/ligand combinations such as FKPB/Shield-1 (rapamycin analog), [26] DHFR/trimethoprim, [27] UnaG/bilirubin [28] or the human estrogen receptor binding domain/tamoxifen, [29] respectively.Avariety of species as well as diverse cellular proteins like kinases,c ell cycle regulatory proteins,t ranscription factors and GTPases have successfully been addressed using DDs. [26,30] Maintaining ac onstant level of the rescuing agent to prevent degradation of the POI can be challenging.T or everse this process and to increase the Organelles and protein aggregates are removed via autophagy,w hereas ad ouble membrane structure encapsulates the substrate and forms the autophagosome. Upon fusion with alysosome the autophagosome becomes an autolysosomei nitiating the degradation process. c) Schematic representation of different Cullin RING ubiquitin ligases (CRLs). CRLs consist of acore protein Cullin (CUL), which is regulated by neddylation, and the RING protein ligase RBX1, interacting with the E2 conjugating enzyme. Substrate specificity is achieved via adaptors and substrate receptors such as Cereblon (CRBN) or the von Hippel-Lindaup rotein (VHL) as well as the E2 ligases. Ub = Ubiquitin. flexibility of the DD system ac onditional FKBP degron was designed which is degraded upon incubation with Shield-1i nstead of its absence ( Figure 2d). [31] Unlike the previous described DDs the ligand-induced degron (LID) can be fused to the C-terminus of the target protein only.

Small Molecule-Assisted Shutoff
Ahighly interesting degron concept is described by small molecule-assisted shutoff (SMASh) which belongs to the family of destabilizing tags.S MASh combines ac onstitutive degron with aprotease and its corresponding cleavage site all derived from the hepatitis Cv irus (Figure 2e). [32] After expression of the SMASh-POI fusion construct the SMASh degron (34 kDa) comprising the constitutive degron is autocleaved from the target protein by the embedded protease. This results in the release of the non-modified, native POI as well as the cleaved SMASh degron which is subsequently degraded due to its constitutive degron. Addition of the hepatitis Cvirus protease inhibitor asunaprevir [33] inhibits the activity of the protease within the SMASh degron resulting in degradation of the whole SMASh-POI fusion construct. Like many previously described degrons the SMASh system can be either fused to the N-or C-terminus of the target protein and has been studied in mammalian cells as well as yeast. [32] Introducing the SMASh-tag to the host genome via CRISPR/Cas creates the endogenous protein (without asunaprevir) or induces its degradation (addition of asunaprevir) in ad ose dependent manner making SMASh ah ighly interesting strategy for target validation. As no existing target protein and only newly synthesized protein is degraded upon asunaprevir addition the SMASh systems can be used to study protein half-life.The time required for depletion of the target protein solely depends on its native half-life.

Light-Activated Degrons
Activating adegron via temperature or asmall molecule is av ery convenient strategy to modulate protein abundance but does not provide aspatial component to control the target protein levels.T oallow spatiotemporal control over the target light-activated degrons (LADs) were designed (Figure 2f). In as traightforward approach auxin was capped using ap hoto cleavable protecting group which can be removed after light irradiation. [34] Am ore sophisticated approach describes the photosensitive degron (20 kDa), atruly light induced degron, which is independent of ligand binding. [35] Illumination induces ac onformational shift in the light sensitive domain exposing an unstructured region, which is degraded by au biquitin-independent mechanism via the proteasome.A similar system, termed blue-light-inducible degron, has been used to rapidly (t 1/2 = 30 min) degrade atarget protein fusion construct in mammalian cells and zebrafish upon blue-light irradiation. [36] Controlling protein levels using light is an elegant and am arginally invasive method which holds great promise for further application. [37] However, depending on the envisioned organism photosensitivity and light penetration might hamper the application of LADs.

TagS pecific Degraders
Avery powerful tool to quickly achieve TPD for aw ide range of target proteins and without the need to identify aspecific binder represents the use of tag specific degraders. In this case,the degrader molecules are optimized to engage the protein tag resulting in effective removal of the tag-POI fusion construct. Thetwo main tags which are widely used as chemical biology tools for this technology are the HaloTag [38] and the dTag system. [39] Compared to many other techniques discussed in this manuscript the mode of action by which the bifunctional degraders designed to degrade HaloTag and FKBP F36V -fusion (dTag system) proteins are working is well characterized and understood (for more detailed information please see section 3.3 PROTACs). In short, the bifunctional degrader brings the tagged protein as well as an E3 ubiquitin ligase in close spatial proximity inducing ubiquitination and subsequent proteasomal degradation of the tagged protein. [3, 5a] TheH aloTag (33 kDa) is aw idely used self-labeling tag introduced by Promega which covalently binds to chloroalkanes. [40] This allows highly selective covalent labeling of the HaloTag protein with as ynthetic ligand of choice.A ttaching an E3 ligase binder to the chloroalkane HaloTag ligands results in efficient and selective degradation of the target protein construct (Figure 3a). [38] HaloTag degraders have been designed to hijack VHL and IAP E3 ligases,t hus far. They have been effective in the low nanomolar range to induce fusion protein degradation in overexpression models as well as in combination with endogenous CRISPR/Cas gene editing.H aloTag degraders have to be applied in stochiometric amounts due to their covalent nature and subsequent removal from the system together with the target protein. This is ag eneral drawback for the HaloTag system as the catalytic turnover presents an attractive feature of PROTACmediated protein degradation.
In contrast, applying sub-stochiometric amounts of the bivalent degrader to achieve full target degradation can be achieved using the dTag system. [39] Thed Taga pproach comprises aF KBP F36V protein tag (12 kDa) fused to the target protein as well as abivalent degrader molecule (dTag). Thed Tags electively binds to FKBP on one site and the E3 ligase Cereblon (CRBN) on the other site and thereby induces degradation of the fusion protein construct (Figure 3b). Thed Taga pproach has already been used multiple times for target validation in various disease models. [41] In contrast to the HaloTag system the dTag can act catalytically and has proven to work for both, in vitro as well as in vivo applications.

Nanobody Induced Target Degradation
Another strategy to achieve selective TPD is based on genetically encoded nanobodies fused to E3 ligases,m ore specifically their substrate recognition proteins (Figure 3c). Ty pically,this technology is used to degrade green fluorescent protein (GFP)-POI fusion constructs using aG FP specific nanobody.N anobodies are small single chain polypeptide antibodies (12-14 kDa) derived from camelid species and can be raised against different antigens.I nitial studies were performed in Drosphila and C. elegans as well as mammalian cells. [42] To establish at imely control over the system the expression of the nanobody-VHL fusion can be achieved by using atetracycline inducible vector system. [42d] Similar results can be obtained using an auxin-dependent nanobody approach where the GFP-targeting nanobody is fused to the AID. [43] Theg enetically encoded degraders (described in section 3.5) resemble the next generation of the anti-GFP nanobody approach. These degraders are not directed against aGFP fusion construct and are directly addressing the target protein itself,t hus allowing for genetically encoded tag free induced degradation.

Degradation of Untagged Target Proteins
Thea dvantage of targeting untagged proteins is that no genetic alterations are required to modify the system in advance.T his setup mimics the pathological state more closely as no tag hampers the function, expression and availability of the target protein and allows to rapidly switch from one cell line to another as well as between organisms or species,r espectively.C onsequently,b eing able to degrade untagged target proteins is pivotal to move towards therapeutic applications.H owever,e verything depends on the availability of as uitable target binder to engage the desired target protein or novel strategies to identify molecular glues.

Hydrophobic Tagging
Maintaining protein homeostasis is essential for the overall wellbeing of the cell. Therefore,u nfolded and misfolded proteins need to be immediately recognized and removed by the cellsq uality control machinery.T his occurs via ap rocess called unfolded protein response (UPR). [44] A characteristic feature of misfolded proteins is the exposure of hydrophobic residues or unstructured domains to the solvent, which is recognized by molecular chaperones.C haperones either rescue the misfolded proteins by supporting their refolding or mediate their degradation by the proteasome in cases refolding is not successful. [45] Am ethod termed hydrophobic tagging (HyT) takes advantage of the UPR machinery by exposing hydrophobic patches on the surface of target proteins thereby triggering their degradation. [46] An adamantyl group or Boc 3 Arg is covalently attached to atarget binder mimicking the partially unfolded protein domain which can induce UPR upon ligand binding (Figure 4a). [3] Although HyT has successfully been used to degrade several target proteins such as the androgen receptor or the pseudokinase Her3, [47] it is predominantly applied as at ool in chemical biology to study UPR in different cellular compartments or to degrade HaloTag fusion proteins. [48] In general HyT suffers from low bioavailability and often shows incomplete target degradation. However,F aslodex (AstraZeneca ), as elective estrogen receptor degrader,i sa pproved for the treatment of hormone-receptor positive breast cancer. [49] Nevertheless,itis important that Faslodex was not designed as aH yT and it is not clear to which extend its therapeutic benefit depends on the induced target degradation.
Sometimes the structural modification turning an inhibitor into adegrader can be very subtle.F or example, BI-3802 induces the degradation of BCL6 in aproteasome dependent manner and significantly outperforms the corresponding inhibitor BI-3812,aclose analog (Figure 5a). [50] Thee xact mode of action for BI-3802 still remains elusive but it does not carry al arge hydrophobic moiety like the typical HyTs.T his poses the questions of how many other supposedly inhibitors are out there that act via protein degradation.

Molecular Glue
Hijacking and reprogramming E3 ligases to induce novel protein-protein interactions (PPIs) and the degradation of neo-substrates-non-native substrates recruited by exogenous ligands-is am ethod already used and optimized by many different viruses to defend themselves against the hosts response. [51] However,itwas still surprising to identify asmall molecule,t halidomide,w hich is able to modulate the substrate scope of the E3 ligase CRBN upon binding. [52] Thalidomide and its close analogs pomalidomide and lenalidomide are part of the immunomodulatory imide drugs (IMiDs) which reprogram CRBN to degrade several zinc finger transcription factors. [6b, 53] IMiDs and other monovalent E3 ligase binders which are able to induce neo-substrate degradation are often referred to as molecular glues (Figure 4b). [6c,54] They induce the interaction of two proteins which otherwise do not show anative affinity for each other. However,w ith the current understanding the design of molecular glues is still hinging on serendipitous discoveries and only very few systems beside the CRBN/IMiD system have been identified thus far. [55] Surprisingly and without intention, while trying to optimize aC RBN based MDM2 bifunctional degrader,s mall structural variations of the molecule lead to the identification of am olecular glue selectively degrading the translation termination factor GSPT1 which has been identified as an relevant off target of heterobifunctional phthalimid based degraders (Figure 5b). [56] Due to their small molecule nature and corresponding drug-like properties molecular glues are highly interesting for therapeutic applications in the field of TPD. [6a,b]

PROTACs
Thet echnique currently creating by far the most excitement in the field of TPD are PROTACs.C ompared to other approaches like TRIM-away (section 3.4) or hydrophobic tagging (section 3.1) PROTACs are highly modular.Aslarge small-molecules they show borderline drug-like properties but have surprisingly proven suitable for oral application. [57] Among all TPD approaches PROTACs are the most rational as ligands for the target protein and the E3 ligase are specifically selected. However,e verything depends on the availability of as uitable target binder.H ighly potent and efficacious PROTACsh ave since been developed against avariety of target proteins such as the bromodomain BRD4, the androgen receptor or Brutonst yrosine kinase. [58] Based on the tremendous success of PROTAC-mediated protein degradation, the first two PROTACsdeveloped by Arvinas , ARV-110 and ARV-471, have entered clinical trials in 2019 targeting the androgen receptor and the estrogen receptor, respectively.F irst results published by Arvinas describing human PK data reveal human exposure in the preclinical efficacious range with long half-lives of 24 hu pt om ultiple days. [59] Initial clinical efficacy data has been presented at the ASCO 2020 annual meeting confirming the PROTACm ode of action in humans.A st he focus of this review is ag eneral overview of various chemical biology techniques to induce TPD more detailed literature on PROTACs can be found elsewhere. [3,5,60] PROTACs are bifunctional molecules comprising atarget binding moiety,a nE 3l igase recruiter and al inker (Figure 4c). Thep redominantly used E3 ligases for PROTACmediated protein degradation are VHL and CRBN with fewer examples harnessing IAP or MDM2. Examples of typical E3 ligase recruiters are shown in Figure 6. PROTACs induce spatial proximity between the target protein and an E3 ligase via the formation of at ernary complex thereby inducing the ubiquitination of the POI. Subsequently,t he ubiquitinated target protein is recognized and degraded by the proteasome.T his event-driven mode of action allows PROTACs to address functions out of reach for traditional small molecule inhibitors and expands the druggable space as biological function can be connected solely to target engagement and is not anymore depending on high affinity target inhibition. [1a,b, 3] However,byengaging the native intracellular protein degradation machinery PROTACsa re limited to intracellular target proteins.T he PROTAC-induced absence of aprotein opens up the target scope to traditionally hard to address non-enzymatic protein families like pseudo kinases or proteins acting as adaptors or scaffolds. [61] Additionally, certain resistance mechanisms can be attenuated or circumvented via degradation of the target protein. [58a,c] Whereas already a10-fold loss in potencydue to mutations of the target protein usually results in insufficient target inhibition (below IC 90 or IC 95 )for small molecules,the residual binding affinity in combination with the subsequent target degradation renders PROTACs still efficacious making them less liable to many frequently observed resistance mutations. [58h, 62] Nevertheless,c ancer cells will still find aw ay to defend themselves against PROTACs.S everal recently published studies have identified mutations of the E3 ligase complex and the signalosome complex as potential resistance mechanisms against different BRD4 degraders. [63] It appears,t hat the acquired mutations in the UPS are either selective to the VHL or CRBN E3 ligase leaving the cell sensitive to the degrader engaging the respective other E3 ligase.
While empiric guidelines like the rule-of-five [64] for the optimization of traditional orally available small molecules inhibitors have been developed and refined over the past decades,o ptimizing the bioavailability for large molecules like PROTAC is still in its infancy. [65] It appears that many predictive assays used for small molecules are less predictive when it comes to PROTACoptimization. On the other hand, due to its catalytic mode of action already alow concentration of PROTAC can be sufficient for effective degradation. Half maximal degradation concentrations (DC 50 s) in the low-and sub nanomolar range are commonly observed even when binding to the E3 ligase or the target protein are only in the micromolar range. [58d, 66] However,m embrane permeability poses as ignificant challenge for molecules of this size and polarity.I narecent study analyzing cell permeability the investigated PROTACs howed ad rastically reduced cellular uptake compared to its parent inhibitor. [67] Despite the reduced uptake,t he PROTAC was still able to degrade its target protein with nanomolar potencyh ighlighting the PROTACs catalytic mode of action. [67,68] Thea bility to perform multiple cycles and act in substoichiometric amounts describes ak ey feature of the PROTACstechnology making the use of covalent target binders very challenging. [69] However,s uccessful covalent Ras G12C and BTK PROTACsh ave been identified, recently. [70] Despite the lack of catalytic turnover covalent PROTACsmight still prove valuable when addressing highly challenging targets.Surprisingly,the use of reversible covalent warheads seems also to have af avorable impact on the cellular uptake of certain PROTACs when targeting ap articular cysteine. [70b,71] Chemoproteomic analysis of many PROTACs revealed their generally excellent selectivity. [61b, 68,72] Furthermore,s tarting from ap romiscuous kinase inhibitor such as foretinib, degraders with significantly improved selectivity profile can be obtained. [66b,72b] Optimizing ab eyond rule-of-five compound with am olecular weight most likely between 700 and 1200 gmol À1 for clinical applications is an enormous challenge as many principles established for small molecule inhibitors do not translate to larger molecules.Despite these challenges,Arvinas has managed to identify orally available PROTACs for their first two clinical programs and other companies claim to have achieved oral bioavailability for their PROTACs,a s well. With only very few design guidelines available PROTACo ptimization is still no straight forward process. [65b] This makes acarefully assembled screening platform invaluable for the characterization of the vast number of PROTACs synthesized during an optimization campaign. Ar epresentative screening tree is outlined in Figure 7. While some assays are straight forward to set up and can be run in an HTS format other assays need to be tediously  or are limited by low throughput. To degrade at arget protein aP ROTACm ust successfully accomplish ac ascade of different steps:i )enter the cell ii)engage the target protein and the E3 ligase iii)form an active ternary complex with the target protein and the E3 ligase iv) ubiquitinate the target protein v) degrade the target protein. [73] While all these processes can be interrogated in individual assays not all of them are adding any value when you are looking to design your first PROTACf or aP OI. From our perspective,t he most important and feasible assays are biochemical and cellular target protein as well as E3 ligase engagement, and most of all target degradation. Consequently,i na ni ndustrial setting these assays need to be available in ah igh throughput format to be able to assess several hundred or more PROTACs.A sm ost PROTAC programs so far are based on established small molecule targets,atarget protein assay (inhibition or binding) is usually available or can be set up easily.The same is true for the most commonly used E3 ligases such as von VHL, CRBN,MDM2 or IAP for which different assays to monitor target engagement are commercially available. [74] Comparing target engagement in live cells with in vitro or lysed conditions allows an evaluation of cellular uptake.S everal published degraders have been evaluated in these assays and can be used as references.However,the most important challenge is to develop asuitable high throughput assay monitoring target protein degradation. While western blotting (WB) is the method of choice used to determine protein levels in almost all PROTAC publications the throughput is too low to evaluate al arge number of molecules at various time points. Nevertheless,WBinadose dependent manner should always be performed as asecondary assay to verify degradation as it is the most specific method using unmodified endogenous protein levels.A sah igh throughput compatible degradation assay several options are amendable with each assay offering opportunities and pitfalls.Using an overexpressed and tagged construct of the target proteins allows easy monitoring (e.g. GFP tagged proteins), quick setup and time resolved degradation in ahigh throughput format but due to potential high expression levels weak degraders might be missed or false positive PROTACs identified which only initiate degradation of the tag instead of the target protein itself. [75] CRISPR/Cas modifications of the endogenous target protein adding asmall tag (HA-tag or NanoLuc )are more time consuming but can increase sensitivity due to endogenous protein levels and am ore favorable target protein/E3 ligase ratio.I mmunofluorescence imaging is av aluable alternative to detect protein degradation but relies on the availability of aspecific antibody and requires ac onstant supply of consumables.I n cell western, TR-FRET or AlphaLISA can be considered as well. To our knowledge ageneral strategy for monitoring high throughput protein degradation has not yet manifested with methods varying for each target protein depending on available tools.N evertheless,h its from the high throughput degradation assays need to be validated in an orthogonal assay such as WB or capillary electrophoresis.F or af ull characterization of the degradersd ownstream effects thorough proteomic analysis is inevitable.I nt his regard, chemoproteomics are emerging as the new workhorse and method of choice to characterize protein degraders as an unbiased approach is important. As has been shown previously,e ven weak binders can yield efficacious degraders and especially for CRBN based PROTACs all effected proteins need to be identified in order to monitor off-targets and mitigate safety concerns. [66b,76] In case no active PROTACs can be identified using this setup or as additional support for rational PROTACo ptimization further assays are useful. Especially monitoring the formation of the ternary complex comprising the target protein, the PROTACa nd the E3 ligase is important at this point. Measuring ternary complex formation has been mainly done by using biophysical assays such as isothermal titration calorimetry (ITC) or surface plasmon resonance (SPR) which are both limited by low or medium throughput, respectively. [77] On the other hand, the information obtained from these measurements is highly valuable and allows to assess cooperative binding, ak ey feature for many highly active PROTACs.
[77a] Biochemical assays like time-resolved fluorescence resonance energy transfer (TR-TRET) or AlphaLISA can be used as well to assess ternary complex formation and provide ah igher throughput compared to ITC or SPR. [72a,77c] If ternary complex formation is evident, ubiquitination of the target protein can be checked. Suitable assays are provided by Promega or can be run via tandem ubiquitin binding entity (TUBE) precipitation [78] followed by WB detection of the target protein. However,t hese assays are usually only run in cases where no degradation can be observed and the PROTACn eeds to be carefully characterized. If aw orking PROTACi si dentified and degradation via the proteasome verified (addition of proteasome inhibitor as control) aubiquitination assay becomes redundant.

Reviews
To aid PROTACoptimization, attempts to crystallize the ternary complex in order to understand the orientation of the two ligands as well as the linker is highly recommended. Structural insight can be tremendously helpful to guide PROTACd esign and can lead to surprising but useful tweaks not obvious at first glance. [61b,77a, 79] Computational modeling can give additional guidance when it comes to PROTACand linker design and should be applied as early as possible during optimization. However,w ithout structural guidance de novo prediction of the ternary complex is highly challenging due to the myriad of possible conformations. While initial PROTAClinkers usually comprise polyethylene glycol or alkyl chains to allow sufficient flexibility for ternary complex formation, the long-term goal is to dial in rigidity in order to increase molecular recognition and overall stability. This process has been nicely demonstrated for the SMARCA2/4 PROTACs. [61b] While several assays have been described for the in vitro optimization of PROTACsv ery little is known about how to design PROTACs for in vivo applications.D espite various xenograft studies using PROTACs are reported, optimization strategies for in vivo experiments are only beginning to emerge. [58c,d, 80] Ther ecently published study on improving bioavailability of ap reviously unstable Brutonst yrosine kinase PROTACtoyield acompound with reduced clearance, increased half-life and exposure is the beginning of hopefully many more examples to follow. [81] Thus far,j udging from alimited amount of data, it appears to be very challenging to establish an in vitro/ in vivo correlation for PROTACs. Without reliable prediction tools available,m any PROTACs will need to be profiled in vivo to optimize bioavailability and efficacyinorder to develop ac linical candidate.

Light Controlled Targeted Protein Degradation
Despite the excitement about PROTACs as an ovel therapeutic modality,PROTACs bear also ahuge potential as achemical biology tool (e.g.section 2.2). Thedesign of photoswitchable PROTACsa llows for temporal control of TPD ( Figure 8a). [82] These PHOTACs [82a] or PhotoPROTACs [82b] can be switched between ab iologically active and inactive state using different wavelength of visible light. In an additional approach to gain temporal control of protein degradation, photocaged PROTACs have been described (Figure 8b). [83] Upon irradiation with ac ertain wavelength the free PROTACisr eleased and degradation is induced. In further studies these light controlled PROTACs might be used to monitor TPD in distinct cellular compartments.

Extracellular Targeted Protein Degradation
Complementary to PROTACs,lysosome targeting chimeras (LYTACs) allow the degradation of extracellular and membrane target proteins. [84] AL YTAC is ab ifunctional molecule comprising an antibody against an extracellular target protein which is fused to an agonistic glycopeptide ligand for the cation-independent mannose-6-phosphate receptor (CI-M6PR) (Figure 8c). TheC I-M6PR internalizes upon ligand binding and transfers its cargo to lysosomal degradation while the receptor itself is recycled and shuttled back to the cell membrane.F irst proof-of-concept studies demonstrate the successful degradation of epidermal growth factor receptor (EGFR), programmed-death-ligand 1( PD-L1) and Apolipoprotein E4. However,a tt his point further studies are needed to assess the potential of this technology and any potential therapeutic application.

Autophagy Mediated Targeted Protein Degradation
Thes econd major pathway of protein degradationautophagy-is pivotal for the degradation of misfolded proteins,a ggregates and especially whole cell organelles.I n ar ecent study the autophagy pathway has been hijacked for the first time to rationally degrade adesired protein as well as fragmented mitochondria (Figure 8d). [85] Based on an (S)guanylate derivative as ar ecognition tag for the autophagosome autophagy targeting chimeras (AUTACs) have been designed to remove cytosolic proteins and organelle targets. Due to their mode of action AUTACs will be limited to cytosolic targets and most likely not be able to act in ac atalytic manner.I nt hat sense,a utophagy mediated TPD can be more precisely perceived as cargo loading onto the autophagosome and subsequent degradation than catalytic turnover as observed for molecular glues or PROTACs.

TRIM-Away
In contrast to PROTACswhich are currently investigated as potential remedies for diseases the TRIM-away technology has agreat potential to be avaluable tool in chemical biology and target validation. [86] TRIM-away harnesses the endogenous ubiquitin ligase TRIM21 which recognizes the constant Fc-region of antibodies ( Figure 9a). Thus,proteins of interest can be targeted by ac onventional antibody which upon TRIM21 recognition induces rapidly ubiquitination and degradation of the protein-antibody complex via the proteasome.I nc ontrast to the catalytic nature of PROTACs the whole complex of target protein, antibody as well as TRIM21 is degraded resulting in consumption of the antibody and the ubiquitin ligase.C onsequently,e ndogenous TRIM21 levels might not be sufficient to achieve complete target degradation which demands TRIM21 supplementation via overexpression or external addition. Forthe successful application TRIM-away requires as elective antibody against the POI which recognizes the native conformation of the target protein. Additionally,t he antibody and supplemented TRIM21 must be delivered via microinjection or electroporation hampering its general applicability.T RIM-away is particularly useful for target validation if DNA-or RNAbased depletion methods have not been successful and can be also applied to non-dividing primary cells for which conventional loss-of-function techniques are particularly challenging.

Genetically Encoded Degrader
As already mentioned above,t he rational design of at argeted protein degrader relies on the availability of as uitable target binder.H owever,m ost proteins of the undruggable proteome are termed undruggable due to the lack of chemical matter.B iologics on the other hand can be raised against most protein targets but are limited to extracellular or surface proteins.N evertheless,b yf using ap eptide or antibody-mimetic directly to ah uman or bacterial E3 ligase so called bioPROTACsw ere designed to specifically degrade challenging intracellular proteins (Fig-ure 9b). [87] As only af ew E3 ligase binders have been identified thus far,P ROTACs are mainly harnessing the four above mentioned E3 ligases of the more than 600 available human E3 ligases.AsbioPROTACs are engineered genetic constructs the whole repertoire of ubiquitin ligases can be utilized for target degradation. In ap roof of concept study,t he protein levels of the oncology target, proliferating cell nuclear antigen (PCNA), could be significantly reduced with 9o ut of 16 tested E3 ligases and much faster than comparable RNAi-based approached would allow. [87a] Additionally,n anobodies fused to several F-box proteins were screened for degradation of activated RhoB(GTP) to study its

Angewandte Chemie
Reviews cellular function. [87d] Although bioPROTACsc an turn into powerful tools for interrogating the degradability of as ubstrate and guiding the selection of the best suited E3 ligase they will be most likely limited to chemical biology applications due to their dependence of genetic encoding.

Conclusion and Outlook
As highlighted by the various achievements within recent years,T PD has matured from ac hemical biology niche to af ully established application for target validation and discovery.F urthermore,T PD in form of molecular glues is established on the market with billions of revenues created by the IMiDs alone and has even advanced to the clinic in the very young field of PROTAC-mediated degradation in 2019. [57,59] Despite the tremendous success TPD is still in its infancythus far only scratching the surface of its full potential especially with respect to expanding the druggable space. Piggybacking on the advances in genome editing technologies (e.g. CRISPR/Cas) as well as chemical proteomics TPD will further flourish as ap owerful strategy in chemical biology foremost in the area of target validation and target mining. Within the near future we will see TPD develop as one of the standard procedures to verify hits identified via genome editing screens.With more TPD approaches already allowing an in vivo application target validation will raise to ac ompletely new level. [41b] With the myriad of methods to choose from one can easily be overwhelmed and struggle to find the best system for the problem at hand. Foracondensed overview of the discussed techniques we summarized their key features in Table 1. However,judging from amedicinal chemistsperspective the most powerful chemical biology tools for target validation appear to be the SMASh-tag and the dTag approach. [32,39] The SMASh-tag requires only asingle genetic modification of the host genome and either produces the native,untagged POI, as without addition of the inhibitor asunaprevir the active protease cleaves the degradation tag from the POI, or the SMASh-POI fusion construct is degraded upon asunaprevir addition. Consequently,t he studied system resembles the native state as closely as possible without am assive tag influencing the availability,a ctivity or general properties of the target protein. Although, in vivo applications for the SMASh-tag have thus far not been reported the small molecule degradation inducer asunaprevir is an approved drug and is predicted to be applied at dosages that are nontoxic in vivo. [32] As aminor drawback, the SMASh-tag can only control the fate of newly synthesized protein and has no effect on already expressed (and degron cleaved) protein levels.Therefore,the clearance of the POI from the system is solely dependent on the protein half-life and might be not suitable for proteins with av ery long half-life.D espite the enormous potential of the SMASh-tag technology only afew examples have been reported to date and it remains to be seen whether SMASh will gain broad acceptance.A na lternative presents the dTag system which rapidly degrades the FKBP F36V -POI fusion proteins from the system. [39] Like the SMASh-tag the dTag only requires as ingle genetic insertion and has already proven to be effective in vivo. [41b, 88] However, the size and attachment to either N-or C-terminus of the FKBP F36V -tag (12 kDa) might influence the target protein in an unfavorable way and needs to be checked beforehand. Future advances in gene editing technologies will improve the ability to insert these tags on an endogenous level making them hopefully available to almost any desired cell line in the future.
While the outlook for the chemical biology side of TPD is highly encouraging it is equally bright for the therapeutic angle targeting untagged proteins.Already in the short period of rationally designing protein degraders,p ost elucidation of the IMiD mode of action (2010), several CRBN modulators with different selectivity profiles haven been identified and three are currently evaluated in clinical trials. [6b,89] This created an excitement towards the identification of additional molecular glue systems with the first results being reported recently. [6a, 54b] Besides molecular glues,P ROTACs have entered the stage and rapidly progressed towards the first clinical trials. [57,59] PROTACsa llow an even more rational approach to target almost any protein for degradation, provided as uitable binder can be identified. However,m ost PROTACprograms reported thus far are based on validated clinical targets.T or educe the risks when exploring an ew clinical modality this cautious initial approach is reasonable Figure 9. Schematic representation of TRIM-Away and bioPROTACs. a) TRIM21 selectively recognizes the Fc-region of an antibody which results in proteasomal degradation of the POI, the antibody as well as TRIM21. b) bioPROTACs are fused to the E3 receptorp roteins and bind to the POI via apeptide or protein recognition domain. Subsequently,the POI gets ubiquitinated and degraded by the proteasome. POI = protein of interest;U b= Ubiquitin. but to unleash the full potential of PROTACsand expand the druggable space anew way of thinking in drug discovery must emerge.B yi nducing degradation of the target protein degraders can exhibit ad ifferential biology compared to inhibitors and address scaffolding roles as well as enzymatic functions.T his provides ac hance as well as ap itfall for the early hit-to-lead process.I ne stablished hit-to-lead processes compounds are often excluded due to assumed off-target effects without any follow up investigation if their cellular effects are stronger than their biochemical IC 50 sw ould predict. With respect to TPD,t his process most likely has removed several potent protein degraders without ever realizing their existence.C onsequently,s ooner or later the hit-to-lead process in industry will have to be adapted accordingly.I nt he future,a iming for the identification of small-molecule target protein degraders instead of inhibitors will emerge as ac ommon strategy in early research.
Considering the recent successes of TPD and especially PROTACs experts in the field are excited to see whether PROTACs can live up to their high expectations.A lthough the first human PK data look promising PROTACsstill have along way to go until they will achieve approval and an even longer way to see how they will perform on the market. Crucial questions like:" How quickly will we see resistance emerge?" or:"What will the main resistance mechanism be?" are waiting to be answered. Efforts to pinpoint common PROTACr esistance mechanisms consistently found mutations hampering the UPS while leaving the target protein unscathed. [63] Nevertheless,o nly long-term clinical use will provide answers.
Despite these open questions the hunt to identify the next clinical candidates is in full swing. Almost every pharmaceutical company is pursuing TPD programs either alone or in collaboration with one of the young biotech companies in the field. Upcoming decisions on several patent applications are monitored closely by everyone working on TPD.While some companies are aiming to discover new chemical matter for the established E3 ligases others are focusing on unraveling the technology itself.However,aset of rules on how to design the perfect PROTACf or ag iven target remains to be crafted. Thus far,the correlation between in vitro and in vivo data still poses amassive challenge when it comes to PROTACdesign. Additionally,i mmense efforts are underway to identify and hijack tissue or disease specific E3 ligases in order to address certain diseases more selectively or increase the therapeutic window.W ithout any doubt many things are happening [a] Despite Faslodex being an approved drug it was not designed as an HyT.

Angewandte Chemie
Reviews around TPD and new ideas and directions are appearing on ac onstant basis.I nspired by the non-native PPIs commonly observed in TPD,i deas are emerging to induce other post translational modifications besides ubiquitination. Early examples recruiting phosphatases or RNAses to eliminate proteins before they are produced will only be the tip of the iceberg. [90] In light of the success of biologics and the emerging progress of nucleotide-based therapeutics as well as the upcoming gene therapies,s mall molecule drug discovery has been looking for its own shot to address the undruggable proteome.TPD is among the technologies which helps to fill this void offering apath towards amedicinal chemistry based approach able to compete with many of the new modalities emerging within the last decades. [91] Within the industry ag oldrush-like atmosphere developed, and it remains to be seen if TPD can live up to its high expectations.H owever, these are definitely great and exciting times to work on TPD and be able to actively shape the next generation of small molecule drug discovery.