E3 Ligase Ligands in Successful PROTACs: An Overview of Syntheses and Linker Attachment Points

Proteolysis-targeting chimeras (PROTACs) have received tremendous attention as a new and exciting class of therapeutic agents that promise to significantly impact drug discovery. These bifunctional molecules consist of a target binding unit, a linker, and an E3 ligase binding moiety. The chemically-induced formation of ternary complexes leads to ubiquitination and proteasomal degradation of target proteins. Among the plethora of E3 ligases, only a few have been utilized for the novel PROTAC technology. However, extensive knowledge on the preparation of E3 ligands and their utilization for PROTACs has already been acquired. This review provides an in-depth analysis of synthetic entries to functionalized ligands for the most relevant E3 ligase ligands, i.e. CRBN, VHL, IAP, and MDM2. Less commonly used E3 ligase and their ligands are also presented. We compare different preparative routes to E3 ligands with respect to feasibility and productivity. A particular focus was set on the chemistry of the linker attachment by discussing the synthetic opportunities to connect the E3 ligand at an appropriate exit vector with a linker to assemble the final PROTAC. This comprehensive review includes many facets involved in the synthesis of such complex molecules and is expected to serve as a compendium to support future synthetic attempts towards PROTACs.


INTRODUCTION
The ubiquitin-proteasome system (UPS) plays a cardinal role in maintaining intracellular protein homeostasis by eliminating misfolded, damaged, and worn-out proteins (Amm et al., 2014). This process consists of a cascade of distinct steps, starting with ubiquitin activation by enzyme E1. Ubiquitin is then passed to the E2 or ubiquitin-conjugating enzyme by trans-thioesterification. Subsequently, E3 ubiquitin ligase promotes the transfer of ubiquitin onto a lysine of the substrate protein. Ubiquitin's own internal lysine residues allow binding of additional ubiquitins, resulting in polyubiquitin tags, which serve as a signal for protein degradation via the 26S proteasome (Kleiger and Mayor, 2014).
Hijacking the UPS and utilizing its functions to degrade the selected protein of interest (POI) has been made possible by proteolysis-targeting chimeras (PROTACs) (Burslem and Crews, 2020) ( Figure 1). These hetero-bifunctional molecules are composed of a POI ligand connected to an E3 ubiquitin ligase ligand by a linker (Figure 1) (Pettersson and Crews, 2019). A functional PROTAC instigates the formation of a ternary complex POI-PROTAC-E3 ligase, which results in the ubiquitination of the POI, followed by proteasomal degradation (Scheepstra et al., 2019). This new modality began accumulating recognition and significance in medicinal chemistry since 2001 when the first proof-of-concept experiments were published (Sakamoto et al., 2001;Burslem and Crews, 2020).

E3 LIGASES
The human genome includes two members of the E1 enzyme family, roughly 40 E2s, and more than 600 E3 ubiquitin ligases (Kleiger and Mayor, 2014). E3 ligases represent a crucial element in protein ubiquitination due to their role in substrate selection and modulation of the cascade's efficiency (Buetow and Huang, 2016;Zheng and Shabek, 2017). They are categorized into three classes, based on their mechanism of ubiquitin transfer. The first and the most abundant class includes approximately 600 RING (Really Interesting New Gene) E3 ligases, which catalyze the direct transfer of ubiquitin from E2 to a substrate. In contrast, the less represented E3 classes HECT (Homologous to E6AP C-terminus) and RBR (RING-between-RING) form a thioester intermediate with ubiquitin via a catalytic cysteine before the transfer to the substrate protein (Buetow and Huang, 2016). Although our understanding of substrate recognition and regulation of ubiquitination is incomplete, the genome's selection of roughly 600 E3 ligases is capable of ubiquitinating a much larger number of protein substrates in a controlled manner with ample specificity (Fisher and Phillips, 2018).
Despite the vast selection of known E3 ligases, only a handful have been successfully utilized in PROTAC compounds (Burslem and Crews, 2020). Following the first utilization of a poorly permeable phosphopeptide moiety to hijack Skp1-Cullin-F box complex (SCF β-TRCP ) to degrade methionine aminopeptidase-2 (Sakamoto et al., 2001), and targeting the von Hippel-Lindau (VHL) tumor suppressor protein with a seven amino acid long sequence ALAPYIP (Schneekloth, et al., 2004), the field has evolved tremendously, resulting in numerous small-molecule E3 ligands, that allow for the development of cellpermeable and biologically active PROTACs (Sun X. et al., 2019). The first of its kind was a PROTAC targeting the androgen receptor, using nutlin ( Figure 8) to recruit the mouse double minute 2 homologue (MDM2) E3 ligase (Schneekloth et al., 2008). Following that, the number of successfully degraded targets using various E3 ligases, such as cellular inhibitor of apoptosis (cIAP) (Itoh et al., 2010), VHL, and cereblon (CRBN) (Sun X. et al., 2019), steeply increased. More recently, additional E3s were explored and used successfully in degraders, i.e., RING-type zinc-finger protein 114 (RNF114) (Spradlin et al., 2019), damage-specific DNA binding protein 1 (DDB1)-CUL4 associated factor 16 (DCAF16) , and Kelchlike ECH-associated protein 1 (KEAP1) (Tong et al., 2020a). However, the majority of recently reported PROTACs still utilize either VHL or CRBN as E3 ligases (Burslem and Crews, 2020); a fact that is corroborated by a high number of different synthetic approaches to obtain these PROTAC building blocks.
Various aspects of degraders have been extensively reviewed in the scientific literature in recent years (Toure and Crews, 2016;Lai and Crews, 2017;An and Fu, 2018;Sun X. et al., 2019;Pettersson and Crews, 2019;Schapira et al., 2019). However, a thorough overview of synthetic efforts leading to the most commonly used ligands for E3 ligases has not been done. Therefore, in this review, we focus on E3 ligase ligands utilized in successful PROTACs. More precisely, we overview the synthetic routes to obtain the E3 ligands and illustrate the possible linker attachment points and types of bonds used to connect the ligands with linkers. The preparation of specific building blocks was reported, as expected, in many publications. However, if no yields were reported or the authors only referred to the original or previously described work, these publications were not referenced in this paper. In addition, for the most commonly used ligases, a statistical overview of the prevalence of E3 ligase ligands and linker attachment options utilized in PROTACs is provided, along with highlighting the contributions of these building blocks to the physicochemical properties of final PROTAC molecules. This review provides the reader with a concise picture of the current state and enables all newcomers to the field a quick go-to-guide in terms of synthetic access to PROTAC building blocks. We hope that this thorough overview will aid in future successful contributions in the protein degradation field.
To date, CRBN has been successfully utilized as the E3 ligase in PROTAC targeting more than 30 different proteins, ranging from those involved in various cancers (Sun X. et al., 2019) and immune disorders (Bassi et al., 2018), to neurodegenerative disease-associated protein Tau (Silva et al., 2019), and even hepatitis C virus protein NS3 (de Wispelaere et al., 2019). The collection of CRBN ligands with different linker attachment options are presented in

Pomalidomide-Based Ligands
We categorized the possible synthetic routes based on the common phthalic anhydride precursor, as most syntheses start from either 3-fluorophthalic anhydride, which is then subjected to condensation with the glutarimide ring and subsequent nucleophilic substitution by a linker with a primary amine, or 3-nitrophtalic anhydride, which is subsequently reduced to pomalidomide.

3-Fluorophthalic Anhydride as a Precursor for Pomalidomide-Based Derivatives
Several options are available to obtain pomalidomide-based PROTAC precursors when 3-fluorophthalic anhydride (4) is used as the main synthon. The glutarimide subunit can be incorporated into 4-fluorothalidomide (5) by using compound 2, which can be formed easily by converting Boc-Gln-OH (1) into 2 via an intramolecular coupling (Scheme 1, steps a-b) (Steinebach et al., 2018). Another option to afford the desired precursor 2 over three steps with a 57% yield was presented where L-glutamine was used as starting material (Varala and Adapa, 2005). Alternatively, 3-aminopiperidine-2,6-dione hydrochloride can be used in place of 2 (Zhou et al., 2018;An et al., 2019). The 3fluorophthalic anhydride (4) is usually used as a commercially available building block. However, it can be easily prepared in high yield by refluxing 3-fluorophthalic acid (3) in acetic anhydride (Zhou et al., 2018). Following condensation of the glutarimide subunit with 3-fluorophthalic anhydride (4), 4fluorothalidomide (5) was obtained in a higher yield by using NaOAc in AcOH under reflux conditions (Steinebach et al., 2018;Zhou et al., 2018), rather than by a method using Et 3 N in THF at 80°C  (Scheme 1, step d). An alternative synthesis towards 5 was reported where L-glutamine was reacted with 3fluorophthalic acid (3) to form 6, followed by a CDI-mediated intramolecular cyclization (Scheme 1, steps e-f). However, the desired product 5 was obtained in an approximately 14% overall yield using this approach (Lu et al., 2015). Compound 5 then allows for simple linker introduction using primary amines and DIPEA in DMF at 90°C, leading to alkylated pomalidomide derivatives 7 (Scheme 1, step g) (Lu et al., 2015;Steinebach et al., 2018;An et al., 2019). It was reported to replace DMF with DMSO for the linker attachment step because of the thermal decomposition of DMF at high temperatures in the presence of a tertiary amine, forming dimethylamine, which can result in the formation of the undesired 4-(dimethylamino)-thalidomide (Steinebach et al., 2018;Steinebach et al., 2019). Recent advances showed that performing the nucleophilic aromatic substitution of compound 5 with primary or secondary amines at elevated temperatures (130°C) generally resulted in a higher yield of desired pomalidomide derivatives (Brownsey et al., 2021).
Numerous studies include thalidomide derivatives with N-alkylated glutarimide ring (e.g., compounds 9 and 10, Scheme 2) as negative controls since they are incapable of binding to CRBN (Buhimschi et al., 2018). Two options are presented for synthesizing such negative controls, the first being the alkylation of glutarimide moiety 8 before conjugation into the final 4-fluorothalidomide (9) (Steinebach et al., 2018). Alternatively, the imide nitrogen of 5 can be alkylated after the condensation of glutarimide and phthalimide parts , or a methyl group can be introduced via Mitsunobu reaction (Steinebach et al., 2018). A significant number of reported PROTACs incorporate a triazole fragment (e.g., compound 13, Scheme 3; compound 34, Scheme 7; compound 39, Scheme 8) as a result of utilizing click reactions between azides and alkynes (e.g., compound 12 in Scheme 3), under conditions for a typical copper-catalyzed Huisgen 1,3-dipolar cycloaddition. Deemed the 'privileged scaffold for PROTACs', triazoles represent numerous advantages since they are easily accessible in high yields under mild reaction conditions, which are highly compatible with other functional groups (Xia et al., 2019).

Paper
Reagents and conditions Yield The route towards pomalidomide derivatives with a twocarbon spacer was nicely elaborated (Zhou et al., 2018) (Scheme 5). Treating 3-nitrophthalic anhydride (15) with benzyl alcohol and benzyl bromide, reducing the nitro group with stannous chloride, and alkylating the resulting amine group with tert-butyl bromoacetate yielded compound 25. This intermediate was then condensed with 3-aminopiperidine-2,6dione hydrochloride into an N-alkylated pomalidomide derivative 26, containing a two-carbon spacer, that allows the attachment of primary amine linkers via the formation of an amide bond (Scheme 5, steps a-e) (Zhou et al., 2018). Alternatively, the synthesis of 26 was described by treating 4-fluorothalidomide (5) with tert-butyl 2-aminoacetate and cleaving the protecting ester with TFA in a 68% overall yield (Scheme 5, step f) (Powell et al., 2018).

Linker Attachment to Pomalidomide
Coupling pomalidomide (17) with the desired linkers was described by numerous authors, utilizing various acyl chloridebearing linkers in THF under reflux for a various amount of time (Scheme 6). The exact conditions and reported yields are collected in Table 2 and may provide a better understanding of the achievable yield range (Lai et al., 2016;Buhimschi et al., 2018;Li et al., 2018;Rana et al., 2019, 6). It should be noted that a side reaction can occur, i.e., acylation of the imide nitrogen as described (Man et al., 2003). In contrast, alkylation of pomalidomide with alkyl halides is considered to be an inferior strategy for linker attachment due to the low yield and poor chemoselectivity of the reaction (Brownsey et al., 2021).
As an alternative approach to N-acylated derivatives, pomalidomide (17) was reacted with bromoacetyl chloride to obtain 29 or chloroacetyl chloride to obtain 30. Compounds 29 and 30 were then refluxed with NaN 3 in acetone overnight to form azide 31 in a 84% (Chen et al., 2019) and 76%  yield over two steps. The azide was then reduced to amine 32, which presents an attachment point for carboxylic acid linkers via amide bond formation (Scheme 7, steps e-f) (Chen et al., 2019). On the other hand, azide 31 was also subjected to click reaction conditions together with a propargyl linker-POI ligand conjugate to form a triazole ring and final PROTAC compounds of type 34 (Scheme 7, step g) .

5-Aminothalidomide Derivatives
Derivatives of 5-aminothalidomide are less commonly utilized in PROTACs, despite that this substitution pattern still presents a valid option for targeting CRBN (Sun X. et al., 2019). Reagents, conditions and yields for the synthesis of 5-fluorothalidomide (36) are comparable to those in Scheme 1 for the preparation of the 4-fluoro analog. The introduction of primary amine linkers by heating the mixture of 36 and the chosen linker alongside DIPEA to obtain 5-aminothalidomide derivatives 37 has been reported (Scheme 8, step b) (Ishoey et al., 2018). Interestingly, the yields for this aromatic nucleophilic substitution were notably lower in comparison with reactions on a 4-fluoro analog. Using propargylamine as the nucleophile provided compound 38, which again offered a facile option for attaching an azide linker-POI ligand conjugate to form final PROTACs of type 39  (Scheme 8, steps c-d).

4-Hydroxythalidomide Used in In-Cell Self-Assembly CLIPTACs
Intracellular formation of PROTAC molecules is possible by the so-called in-cell self-assembly CLIPTACs, an example of which was described for the degradation of bromodomain-containing protein 4 (BRD4) and extracellular signal-regulated kinase 1/2. In this case, 4-hydroxythalidomide was tagged with tetrazine, while the ligands for the POIs were tagged with trans-cyclo-octene. The combination of the two precursors underwent a bio-orthogonal click reaction to form the active chimera intracellularly. Utilizing this concept might overcome the cellular permeability issues of some PROTACs since the two small precursor molecules have a higher ability to pass through cellular membranes than one large compound (Lebraud et al., 2016).
As presented in Scheme 11, methanolysis and subsequent methylation of 3-hydroxyphthalic anhydride (44) yielded dimethyl ester 47, which was then alkylated under Mitsunobu conditions leading to O-alkylated derivative 48. This represents an alternative to most syntheses, in which the linker attachment is performed only after the thalidomide portion of the molecule is fully assembled. Basic reaction conditions resulted in the hydrolysis of the methyl esters of 48, followed by the condensation with the glutarimide ring and tert-butyl ester cleavage under acidic conditions to obtain 4-O-alkylated thalidomide derivative 49. Amide coupling for the attachment of the tetrazine moiety yielded compound 50, which was finally reacted intracellularly with trans-cyclo-octene, bound to the POI ligand (Lebraud et al., 2016).

4-Hydroxythalidomide Derivatives With a Two-Carbon Spacer
Alkylating the 4-hydroxyl group of 45 with tert-butyl bromoacetate or benzyl glycolate and subsequent removal of the protecting group produces compound 53, a standard building block, containing a flexible 'spacer', which is ready for linker attachment via an amide bond (Scheme 12). The highest yield of over two steps to obtain 53 was reported to be 78% (Remillard et al., 2017). In another study, a yield of only 41% was reached, primarily due to the low conversion rate of Boc-protected derivative 52a to 53 using formic acid (Chessum et al., 2018). A synthesis of 53 was reported by using benzyl glycolate and Mitsunobu conditions, which yielded the desired product 52b in 73% (Lohbeck and Miller, 2016). Coupling reaction yields span between 34 and 85% for a selection of different linkers ( Table 4) (Lohbeck and Miller, 2016;Remillard et al., 2017;Chessum et al., 2018;Zhou et al., 2018, Fischer et al., 2014. Importantly, the selective alkylation of the phenolic group was confirmed by means of HMBC spectra (Lohbeck and Miller, 2016). Alternatively, a 2-chloro-N-acetamide-bearing linker was attached onto phenol 47 to obtain an O-alkylated ester 55. This compound was then first converted to 56, followed by condensation with 3aminopiperidine-2,6-dione to form 54 (Scheme 12, steps d-f). The overall yield of this reaction sequence was approximately 16% (Fischer et al., 2014).

Lenalidomide-Based Ligands
Utilizing lenalidomide-based ligands poses some advantages over using thalidomide and its derivatives to hijack CRBN, as the absence of one phthalimide carbonyl group results in a decreased TPSA, better physicochemical properties, and a higher metabolic and chemical stability (Hoffmann et al., 2013). Additionally, some lenalidomide-based PROTACs displayed a higher level of induced target degradation than their pomalidomide-based counterparts (Qiu et al., 2019). Compound 66 was obtained by bromination of the starting nitrobenzene derivative 65 with N-bromosuccinimide (NBS) in CCl 4 using azobisisobutyronitrile (AIBN) as an initiator of radical bromination, with reported yields of 88% (Balaev et al., 2013) and 49% (Chaulet et al., 2011). An alternative, high-yielding (98%) and green approach for this bromination was presented, where the reaction was carried out in a non-halogenated solvent, i.e., methyl acetate (Ponomaryov et al., 2015). The following condensation with the glutarimide ring was achieved by the addition of a base and heating the solution at 50-55°C, yielding 57% (Et 3 N in MeCN; Chaulet et al., 2011), 86% (Et 3 N in DMF; Balaev et al., 2013), and 89% (K 2 CO 3 in NMP, Ponomaryov et al., 2015) of the desired nitro product 67. Optimal conditions for the subsequent reduction to lenalidomide (68), i.e., a Pd/C-catalyzed hydrogenation, were described (Chaulet et al., 2011). Alternatives include using Pd(OH) 2 (Balaev et al., 2013) or iron-ammonium chloride (Ponomaryov et al., 2015), but both procedures led to the product in a lower yield. Selective derivatization of the 4-amino position of lenalidomide (68) were performed with of bromo or iodo linkers and DIPEA in NMP at 110°C for 12 h to yield derivatives 69 (Scheme 14) (Qiu et al., 2019). Carboxylic acid linkers were attached via an amide bond to form derivatives 70 (Zhang F. et al., 2020).
Frontiers in Chemistry | www.frontiersin.org July 2021 | Volume 9 | Article 707317 these compounds was nicely described recently (Sun Y. et al., 2019). Methyl 3-bromo-2-methylbenzoate (76) was subjected to radical bromination using NBS and AIBN in CHCl 3 to yield 77 (Sun Y. et al., 2019). A higher yield of 90% was reported for a similar radical bromination reaction, where benzene was used as the solvent (Zhou et al., 2018). After condensation with the glutarimide ring to yield bromo-lenalidomide (78), linker attachment was achieved through the Sonogashira cross-coupling reaction to afford compounds 79, with yields spanning between 41 and 81%, depending on the linker used (Sun Y. et al., 2019;Li et al., 2019;Wang et al., 2019). The reduction to 80 was carried out through Pd/C-catalyzed hydrogenation (Sun Y. et al., 2019;Li et al., 2019) (Scheme 16). Alkyl-connected lenalidomide analogs can also be synthesized via the Suzuki cross-coupling reaction (Xiao et al., 2020). The amino group of lenalidomide (68) group was converted into an arylboronic ester 81 through a metal-free pinacol borylation reaction under Sandmeyer-type transformation (Scheme 17). Compound 82 was obtained through oxidative hydrolysis and then joined with tert-butyl bromoacetate by using Pd(PPh 3 ) 4 as a coupling catalyst for Suzuki cross-coupling. Ester hydrolysis afforded compound 83, which is suitable for amine linker attachment to form compounds 84 (Xiao et al., 2020).
It should be mentioned here that synthetic approaches towards hydroxyl analogs of lenalidomide were disclosed recently (Hansen et al., 2020). Although these compounds were not utilized in PROTACs, the syntheses might prove very useful in further research on lenalidomide-derived degraders.

Tricyclic Imide Moiety
The tricyclic imide moiety 86 (Scheme 18) was used in a single PROTAC, which targeted the NS3 protein in virus hepatitis C. Compound 86 had a higher binding affinity for CRBN and did not result in the degradation of neo-substrates, such as IKZF1 and IKZF3 (de Wispelaere et al., 2019). The condensation of 1,8-naphthalic anhydride (85) with the glutarimide ring was performed microwaveassisted (Burslem et al., 2018). 5-Hydroxy derivative 87 was condensed to 88 in a similar way, enabling halogen linker attachment to yield an ether bond-connected linker (Gray et al., 2020).
Frontiers in Chemistry | www.frontiersin.org July 2021 | Volume 9 | Article 707317 incorporating azobenzene photoswitches, with the trans configuration presenting the resting, inactive state. The active cis isomer can be obtained by irradiation with light of specific wavelengths. In one of the cases, the authors incorporated the azobenzene switch directly to lenalidomide (68) to give compound 90. They then derivatized the hydroxyl group to yield 91, which enabled amine linker attachment via an amide bond. Alternatively, compound 53, which contains a flexible two carbon spacer, was coupled with 4,4'-azodianiline to give 93, onto which a POI ligand-linker conjugate with a carboxylic acid was attached to yield 94 (Reynders et al., 2020).

Caged Cereblon Ligands
Apart from PHOTACs, an alternative option that enables the control of the location and timing of targeted proteolysis is incorporating a photocleavable group into a motif that is essential for binding to the E3 ligase (Xue et al., 2019;Naro et al., 2020). The imide moiety of thalidomide's glutarimide ring thus presents an ideal position for attaching a photolabile moiety, such as the nitroveratryloxy-carbonyl (Xue et al., 2019) or 6nitropiperonyloxymethyl (NPOM) group (Naro et al., 2020). In the former study, the imide moiety of starting material 95 was derivatized to form 96 prior to the linker and POI ligand attachment (Xue et al., 2019), while in the latter study, the NPOM group was attached to a conjugate of 4hydroxythalidomide and a linker (54) to form 98. The POI ligand was then coupled via an amide bond to obtain final PROTAC 99 (Naro et al., 2020) (Scheme 20).
Frontiers in Chemistry | www.frontiersin.org July 2021 | Volume 9 | Article 707317 regions. Accordingly, positions that could be derivatized without negatively affecting the critical affinity were identified (Bondeson et al., 2015;Buckley et al., 2015;Zengerle et al., 2015;Maniaci et al., 2017). These include connection via an amide bond after the amino acid tert-leucine (A), phenolic linkage point at the benzene ring (B), link via a thioether at the left-hand side amino acid (C), and via the benzylic methylene group (D) (Figure 4).
A: Connection via an Amide Bond after tert-Leucine von Hippel-Lindau Ligand 1 The key intermediate for the synthesis of VHL ligand 107 (i.e., VHL A1) is compound 105, which can be formed by using a Pd-catalyzed arylation of 4-bromobenzonitrile 103 and subsequent reduction of the nitrile group of 104, for which an array of methods with varying yields has been published (Buckley et al., 2012a;Galdeano et al., 2014;. LiAlH 4 was used which resulted in the desired product 105 in 63% yield , while a NaBH 4 -CoCl 2 combination led to a 29% conversion at 0°C (Galdeano et al., 2014) and 73% at 4°C (Buckley et al., 2012a). Alternatively, a synthetic strategy was reported comprising the conversion of 4-bromobenzylamine (100) into compound 105 in three steps with an overall yield of 18% (Scheme 21, steps a-c) (Buckley et al., 2012a). With compound 105 in hand, the subsequent reaction steps were very straightforward. For example, standard coupling conditions enabled the formation of an amide bond with Boc-L-hydroxyproline, followed by acid-mediated cleavage of the Boc protecting group, which afforded 106. Finally, an amide bond with Boc-L-tert-leucine was formed, and Boc deprotection of the terminal amine yielded compound 107, which allowed derivatization with carboxylic acid linkers to give conjugates 108 (Scheme 21, steps f-h) (Galdeano et al., 2014;Steinebach et al., 2020a).
Inverting the configuration at the hydroxyproline moiety results in a loss of binding affinity for VHL, and such modified compounds are mostly incorporated into negative control VHL-based PROTACs (Raina et al., 2016). Using N-Boc-cis-4-hydroxy-L-proline (109) in place of Boc-L-hydroxyproline and coupling with 105 yielded the VHL non-binding ligand 111 (Scheme 22) .
coupling 105 with the dipeptide 116, which was prepared from hydroxyproline methyl ester (115) and N-Boc-L-tert-leucine. Incorporating an element of convergent synthesis helped to increase the overall yield (Han et al., 2019).
von Hippel-Lindau Ligand 2 In the course of design and optimization of VHL ligands, an introduction of an (S)-methyl group on the benzylic carbon atom has improved the binding affinity to VHL. Namely, the potency of the methyl-substituted ligand is three times better than of the non-substituted ligand 107 (Han et al., 2019). Synthesis of the key intermediate 120 is analogous to the synthetic route described in Scheme 23, using (S)-(-)-4-bromo-α-methylbenzylamine (117) as starting material. The left-hand side dipeptide moiety could be assembled either by convergent synthesis (Raina et al., 2016) or linear synthesis  to yield 122, which was then ready for attaching carboxylic acid linker-POI ligand conjugates by a coupling reaction to give derivatives 123 (Scheme 24) (Raina et al., 2016;Hu et al., 2019).
An alternative synthetic route used 4-bromo-2hydroxybenzaldehyde (141) in place of 4-bromo-2-hydroxybenzonitrile (126, Scheme 25) (Steinebach et al., 2020a). Starting material 141 is easily accessible by ortho-formylating 3bromophenol (139) or by transforming 4-bromosalicyclic acid (140) into a Weinreb amide and its subsequent reduction (Scheme 26). Compound 143 was obtained through reductive amination of 141 with tert-butyl carbamate under mild conditions and Heck coupling in a higher yield compared to the analogous synthesis of compound 128 (Scheme 25, step c). The phenol group of 143 was then protected to prevent the formation of acylated by-products in the following coupling reactions (Scheme 26, steps f-j). The key intermediate 145 was generated from 144 and Boc-L-hydroxyproline and then coupled with 125 and deprotected to give VHL ligand 130. Alternatively, forming an amide bond between 145 and Boc-L-tert-leucine yielded compound 146, which was then Boc-deprotected and derivatized into VHL ligands 134 and 135 (Steinebach et al., 2020b).

D: Connection via the Benzylic Position
Based on analyses of co-crystal structures of VHL ligand 122 (Scheme 24) in the active site of the enzyme, the (S)-methyl group of the VHL ligand was found to be exposed to the solvent and therefore represents a possible liker attachment point for the design of PROTACs. 4-Methylthiazole was coupled with commercially available 151 to yield 152, to which a desired linker-POI ligand conjugate was attached via an amide bond. Boc deprotection afforded compound 153, and the left-hand side dipeptide part was attached to form conjugates 154 (Scheme 28) (Han et al., 2019).

Caged von Hippel-Lindau Ligands
The concept of caged E3 ligase ligands was used for the controlled degradation of BRD4 by incorporating a photocleavable 4,5-dimethoxy-2-nitro-benzyl group (DMNB), bound to the hydroxyproline core of the VHL ligand, connected via a linker to pan-bromodomain inhibitor JQ1. Following irradiation with a wavelength of 365 nm, the PROTAC could be uncaged, which triggered the degradation of BRD4. To prepare a caged PROTAC, the VHL ligand 122 was first N-Boc protected, followed by the functionalization of hydroxyl group by forming an ether bond with the DMNB group using phase transfer catalysis to yield 164. After Boc deprotection, a carboxylic acid linker was introduced via an amide bond to form 165 (Scheme 30) (Kounde et al., 2020). Additionally, the concept was also utilized for the degradation of estrogen related receptor α, where a diethylamino coumarin (DEACM) group was installed at the hydroxyl group of the VHL ligand via a carbonate linkage. Irradiation with a wavelength of 360 nm causes the photolysis and subsequent decaging of the VHL ligand, thus activating the degrader. Compound 167 was obtained from starting material 166 over two steps and then converted to a chloroformate before being attached to a POI ligand-linker-VHL ligand conjugate, forming the final caged PROTAC 168 (Naro et al., 2020).

Statistical Overview of Utilized von Hippel-Lindau Ligands
Using data extracted from PROTAC-DB ( commonly utilized, at about 4%. In comparison, attachment via a thioether at the left-hand side amino acid could be found in only around 1% of PROTACs ( Figure 5).

Inhibitor of Apoptosis Proteins Ligand A: Bestatin Aromatic α-Aminoaldehydes as a Starting Material for Bestatin Synthesis
Several authors proposed different synthetic routes for the synthesis of bestatin, utilizing aromatic α-aminoaldehydes as starting compounds (Scheme 31). For example, compound 169 was treated with nitromethane to afford a diastereomeric mixture of nitroaldols 170, which were then converted into a mixture of dimethyl oxazolidines, out of which the desired compound 171 was separated by silica gel column chromatography in a 54% yield. Compound 172 was obtained by a Nef reaction and then coupled with L-leucine tert-butyl ester to yield 173. Finally, Boc cleavage using TFA afforded bestatin (179) in an overall yield of 24% (Scheme 30, steps a-e) (Shang et al., 2018). An alternative route started from aldehyde 174, which was converted to syn-aminoalcohol 175 in a 96% yield and a 9.5:1 syn/anti stereochemic ratio. The hydroxyl group was then protected with a Bn group to obtain 176, followed by terminal alkyne oxidation to carboxylic acid 177. Coupling reaction with L-leucine methyl ester afforded compound 178, and removal of the protecting groups led to the desired product 179 with an overall yield of 59% (Scheme 31, steps f-j) (Lee et al., 2003). Furthermore, a one-pot method was described, in which starting materials 180, 181, and 182 were joined into 183 with 63% yield. Following the deprotection, bestatin (179) was obtained in an overall yield of 60% (Scheme 31, steps k-l) (Nemoto et al., 2000).

Alternative Routes for the Synthesis of Bestatin
One route included the treatment of (2-nitroethyl)benzene (184) with ethyl glyoxalate in Shibasaki's asymmetric Henry reaction, which was catalyzed by an optically active lanthanum-(R)binaphthol complex. Compound 185 was then O-acetylated before reducing the nitro group to yield 186. Following N-Boc protection, 187 was coupled with L-leucine benzyl ester, followed by immediate deprotection of the terminal carboxylic moiety.
Both protecting groups of 188 were removed to give bestatin (179) in an overall yield of 26% (Scheme 32) . A procedure partly derived from the patent literature started with the treatment of the Meldrum's acid 189 with phenylacetyl chloride to yield 190, which was then chlorinated using sulfuryl chloride to form 191. The following asymmetric hydrogenation using a rutheniumphosphine complex afforded compound 192, which was then subjected to epoxidation to obtain 193 . Compound 194 was synthesized through an MgBr2mediated ring opening of 193. Treatment with NaN 3 afforded the azide derivative 195, which was then hydrogenated and Boc-protected to give compound 196. Hydrolysis of the methyl ester allowed coupling of 197 with L-leucine, and deprotection of 198 yielded bestatin (179) (Scheme 33) (Righi et al., 2003).

Inhibitor of Apoptosis Proteins Ligand B: MV1 Derivative
A similar stepwise peptide synthesis of IAP ligand B using starting biphenyl 203 was described (Itoh et al., 2012;Shibata et al., 2017b). Coupling with N-Boc-L-proline yielded compound 204, the following coupling with Boc-L-cyclohexylglycine gave 205, and finally adding Boc-Nmethyl-L-alanine afforded 206. Catalytic reduction cleaved the O-benzyl group and allowed for coupling of the resulting 207 (Boc-protected MV1 derivative) with POI ligand-linker amine conjugates, giving bifunctional derivatives 208, which were Boc-deprotected to obtain final SNIPER compounds 209 (Scheme 35) (Itoh et al., 2012;Ohoka et al., 2017b;Shibata et al., 2017b). A solid-phase peptide synthesis for an IAP ligand B derivative on a 2-chlorotrityl chloride resin was reported. The stepwise procedure was performed using HCTU, HOBt and DIPEA for coupling, followed by the addition of 20% piperidine in DMF to remove the Fmoc group after each step (Scheme 36, steps a-d). Finally, 214 was treated with 1% TFA in CH 2 Cl 2 to remove the resin and to obtain 207 (Boc-protected IAP ligand B) .

Inhibitor of Apoptosis Proteins Ligand C: LCL-161 Derivative
The procedure for the synthesis of IAP ligand C Shibata et al., 2017b) is shown in Scheme 37. The starting (tert-butoxycarbonyl)-L-proline (215) was converted into 216 over two steps before building the thiazole fragment to give 217. Following the attachment of the 3-hydroxyphenyl building block to obtain 218, the hydroxyl group was deprotected, and the right-hand side of the molecule was built by coupling with Boc-L-cyclohexylglycine to yield 219, and Boc-N-methyl-L-alanine to produce the Boc-protected IAP ligand C (compound 220) Shibata et al., 2017b). Tosylate-containing POI ligandlinker conjugates were attached to the phenol of 220 by heating in DMF or DMSO using K 2 CO 3 as a base, with yields spanning between 62 and 81%, depending on the conjugate used Shibata et al., 2017b;Shibata et al., 2017a;Shimokawa et al., 2017). Alternatively, POI ligand-linker conjugates with a terminal hydroxyl group were attached under Mitsunobu reaction conditions . Final SNIPER molecules of type 222 were obtained by Boc-deprotection of bifunctional conjugates 221 (Scheme 37) Shibata et al., 2017b;Shimokawa et al., 2017).

Inhibitor of Apoptosis Proteins Ligand D
Building block 227 was synthesized from N-(tert-butoxycarbonyl)-N-methyl-L-alanine (226) as starting compound using EDC/HOBtmediated coupling and subsequent hydrolysis of the methyl ester Mares et al., 2020). Compound 227 was then coupled with 230, which was synthesized by subsequent protection of the hydroxyl group and deprotection of the carboxylic acid of 228 to form (2S,4R)-1-(tert-butoxycarbonyl)-4-(tosyloxy)pyrrolidine-2carboxylic acid (229), followed by the coupling of 2,6difluoroaniline. The tosylate group of 231 was transformed into an azide and then reduced to an amine yielding 232, which was then coupled with a carboxylic acid-containing POI ligand-linker conjugate and Boc-deprotected to yield final SNIPER compounds 233 (Scheme 39) (Anderson et al., 2020, 6;Mares et al., 2020).

Inhibitor of Apoptosis Proteins Ligand E: A410099 Derivative
The synhesis of IAP ligand E is described in patents and consists of a stepwise coupling procedure (Scheme 40). The orthogonally protected 234 was first coupled with (S)-1,2,3,4-tertahydronaphthalen-1-amine to form 235, which was then Boc-deprotected and coupled with Boc-L-cyclohexylglycine. The resulting 236 was Boc-deprotected and coupled with Boc-N-methyl-L-alanine to yield 237. The Fmoc protection of 4-amino group on the proline fragment allowed for the selective deprotection to obtain 238 (Borzilleri et al., 2014;Mischke, 2014). Carboxylic acid-containing POI ligand-linker conjugates were attached and the N-terminal amino group was Boc-deprotected to give final SNIPER compounds 239 (Shah et al., 2020). As an alternative, 238 was coupled with a carboxylic acid-containing linker, the product was Boc-deprotected and the POI ligand was attached to obtain the desired SNIPER compounds (Nunes et al., 2019).
TABLE 8 | Reagents, conditions, and yields for the conversion of 220 to 221 (Scheme 37, step f).

Inhibitor of Apoptosis Proteins Ligand I
IAP ligand I was utilized in PROTACs (Dragovich et al., 2020) and could be synthesized based on the following procedure (Kester et al., 2013). By coupling compound 254 with methyl 4-(chlorocarbonyl)benzoate and the subsequent methyl ester hydrolysis, intermediate 255 was formed. POI ligand-linkeramine conjugates were then coupled, and the obtained products were Boc-deprotected to give the final PROTAC compound 256 (Scheme 44) (Dragovich et al., 2020).
Frontiers in Chemistry | www.frontiersin.org July 2021 | Volume 9 | Article 707317 statistical overview was done to determine the frequency of various IAP ligands and linker attachment options used in PROTAC compounds (Figure 7). LCL-161 derivatives were most commonly utilized, with around 30% of PROTACs incorporating its structure. Following closely was bestatine, with MV1 derivatives and IAP ligand E having a lower presence at 10 and 9%, respectively. Other IAP ligands are less common, with fewer than 3% representation.

MDM2
The p53 protein is a product of the tumor-suppressor gene and acts as a transcription factor that gets activated when cell stress occurs, especially upon the occurrence of DNA damage. Activation of the p53 network results in the inhibition of the cell cycle and can lead to apoptosis of damaged cells to prevent their unhindered growth, thus acting as an important tumor suppressor (Levine, 1997;Vogelstein et al., 2000). The effects of p53 are controlled in an  (Steinebach et al., 2020a).
Frontiers in Chemistry | www.frontiersin.org July 2021 | Volume 9 | Article 707317 autoregulatory negative feedback loop involving MDM2, which belongs to the family of RING finger ubiquitin ligases (Michael and Oren, 2003). p53 induces the expression of MDM2, which in turn leads to the repression of p53 activity through binding of MDM2 to p53 and blocking its function, as well as through MDM2-mediated ubiquitination and subsequent degradation of p53 by the proteasome (Michael and Oren, 2003;Vassilev et al., 2004). Excessive activity of MDM2 has been observed in numerous malignancies, which makes it a promising target for the treatment of tumors due to its dual-mode mechanism of interaction with p53 (Momand et al., 1998). The E3 ligase activity of MDM2 was utilized in the first smallmolecule PROTAC by incorporating a molecule belonging to a class of imidazoline derivatives called nutlins, which bind to MDM2 in a nano-to micromolar range (Vassilev et al., 2004;Schneekloth et al., 2008). Additional proteins to be successfully degraded through MDM2-mediated ubiquitination include BTK (Sun et al., 2018), PARP1 , and TrkC (Zhao and Burgess, 2019), among others. Utilized ligands are collected in Figure 8.
An alternative synthetic strategy was devised (Scheme 47), by attaching the linker to tert-butyl 3-oxopiperazine-1-carboxylate 268 prior to urea bond formation with 267 to give compound 270. Cyclization to 271 was obtained similarly as depicted in Scheme 45 (Nietzold et al., 2019).

RNF114
Gene transcription is regulated in a crucial way by the zincfinger gene family. One of the members is ZNF313, also known as RNF114 or ZNF228, which contains both C 2 H 2 and RING-finger structure. Along with the N-terminal RING domain, it also has a C-terminal ubiquitin-interacting motif (UIM), both of which are responsible for RNF114s E3 ligase activity (Bijlmakers et al., 2011;Han et al., 2013). Nimbolide, a limonoid natural product derived from the Neem tree (Azadirachta indica), was recently found to primarily target RNF114 by covalently modifying its cysteine-8, thus leading to impaired E3 ligase activity of RNF114. As a result, substrate recognition is blocked, which leads to the stabilization of tumor suppressors p21 and p57, explaining nimbolide's antiproliferative effects. By incorporating nimbolide into a BRD4-targeting PROTAC, RNF114 was successfully established as a viable E3 ligase option for targeted protein degradation (Spradlin et al., 2019).

Alternative RNF114 Ligand
A small molecule, that accessed the same cysteine targeted by nimbolide was incorporated into a BRD4-targeting PROTAC (Luo et al., 2021). Starting material 296 was first protected with a tetrahydropyranyl ether to form 297 and then reacted with 4bromoacetophenone. Following a deprotection, 298 was obtained and the OH alkylated with a bromo linker, yielding 299. Finally, compound 300 can be achieved over 2 steps (Scheme 52) (Luo et al., 2021).

DCAF16
DCAF16 is a member of the damage-specific DNA binding protein 1 (DDB1)-CUL4 associated factor (DCAF) protein group, which act as substrate-recognition receptors within the UPS (Liang et al., 2017). It consists of 216 amino acids and contains eight cysteine residues, four of which are clustered together between amino acids 173 and 179. By using broadly reactive, cysteine-directed electrophilic fragments, a successful covalent modification of DCAF16 was achieved to induce the degradation of BRD4 and FKBP12 . The authors suggested that utilizing such electrophilic PROTACs may provide certain advantages in the field of targeted protein degradation, as DCAF16 seems to exclusively promote the degradation of nuclear proteins, sparing cytosolic ones. The covalent interaction between DCAF16 and the chimeric molecule allows for protein degradation at low fractional engagement and could SCHEME 51 | Derivatization of nimbolide (292)  The aromatic amine group underwent a reaction with chloroacetyl chloride, giving compound 307, which was then treated with TFA to remove the Boc-protecting group and enable coupling with primary amine linkers (Scheme 54) .

Electrophilic DCAF16 Binder 3
For the third structural type of DCAF binders (compounds 312), the synthesis started with the methyl ester 309, which was hydrolyzed to form 310 and then coupled with primary amine linkers. Compound 311 was then treated with acryloyl chloride to incorporate a cysteine-targeting moiety into conjugates 312 (Scheme 55) .
Frontiers in Chemistry | www.frontiersin.org July 2021 | Volume 9 | Article 707317 activity, it has been found that indisulam stabilizes the interaction between DCAF15 and an essential splicing factor RBM39, which leads to RBM39 polyubiquitination and proteasomal degradation, thus inhibiting cell growth (Han et al., 2017;Bussiere et al., 2020). Building on the known pharmacological activity of indisulam, its structure was modified to enable linker attachment and incorporation of the E3 ligase ligand into PROTACs (Zoppi et al., 2019). Starting material 313 was reacted with 4formylbenzenesulfonyl chloride to obtain sulfonamide 314. Reductive amination lead to amine 315, which was then coupled with POI ligand-linker-NH 2 conjugates to give PROTACs 316, thus successfully expanding the E3 ligase options for targeted protein degradation (Scheme 56). However, activity of such PROTACs was only moderate (Zoppi et al., 2019).

KEAP1
Kelch-like ECH-associated protein 1 (KEAP1) plays a key role in regulating the nuclear factor erythroid 2-related factor 2 (NRF2), which is involved in the cellular cytoprotective response to electrophiles and reactive oxygen species. Being a part of a homodimeric KEAP1/Cul3 complex that possesses E3 ubiquitin ligase activity, KEAP1 works as a substrate receptor and is responsible for selectively recognizing NRF2 and linking it to Cul3 for its ubiquitination (Davies et al., 2016;Jiang et al., 2016). Through forming reversible covalent interactions with cysteines of KEAP1, BRD4 degradation was achieved by recruiting KEAP1/Cul3 E3 ligase activity using the highly electron-deficient cyanoenone moiety-containing triterpene derivative bardoxolone (Tong et al., 2020a). Frontiers in Chemistry | www.frontiersin.org July 2021 | Volume 9 | Article 707317 38

Bardoxolone Derivatives
With oleanolic acid 317 as starting material, an efficient, five-step synthesis of bardoxolone methyl (321) was accomplished (Fu and Gribble, 2013) (Scheme 57). After forming a methyl ester 318, the compound was oxidized to give 319, which was then transformed into 320 over two steps. The final step consisted of substituting the bromo group with a cyano, to form bardoxolone methyl with an overall yield of 50% (Fu and Gribble, 2013). Both bardoxolone methyl and bardoxolone-CO 2 H are commercially available.
Final PROTACs were assembled by using bardoxolone (322) and coupling it with POI ligand-linker-NH 2 conjugates under standard conditions using HATU and DIPEA. Alternatively, bardoxolone was subjected to Pd/C-catalyzed hydrogenation to give compound 324, which was incorporated into negativecontrol compounds 325 with the same coupling procedure (Scheme 58) (Tong et al., 2020a).

FEM1B
FEM1B was recently discovered to play a role in regulating the cellular response to reductive stress, which can lead to various diseases, such as diabetes, cardiomyopathy, or cancer. During a depletion of reactive oxygen species (ROS), FEM1B recognizes reduced cysteines on its target Folliculin-interacting protein 1 (FNIP1) and induces its ubiquitination and subsequent degradation, which restores mitochondrial activity and redox homeostasis of the cell (Manford et al., 2020). The key cysteine residue C186 was noted to present a possible target for developing a FEM1B recruiter useful in the field of targeted protein degradation. Through screening a library of cysteine-reactive covalent ligands, compound EN106 was identified and its structure modified to be incorporated into BRD4-targeting PROTACs (Henning et al., 2021).
An acetate spacer was attached to starting material 326 to give methyl ester 327. After the nitro group was reduced, amine 328 was alkylated with acrylonitrile, followed by acylation with chloroacetyl chloride to yield 329. After hydrolyzing the methyl ester, POI ligandlinker-NH 2 conjugate was attached through coupling to provide PROTAC 330 (Scheme 59) (Henning et al., 2021).

ITE Derivatives
To synthesise endogenous AhR ligand ITE, L-cysteine methyl ester was acylated with starting material 335 to obtain glyoxylamide 336. The following cyclization was performed using TiCl 4 in CH 2 Cl 2 , forming thiazoline ester 337. Oxidation with MnO 2 yielded ITE (338) (DeLuca et al., 2003), the methyl ester of which was then hydrolyzed and available for coupling with amine linkers to obtain compound 339, available for incorporation into chimeric degraders (Scheme 61) .

CONCLUSION
Principles of PROTAC design and tackling the challenges that accompany the field were explored extensively (Maple et al., 2019). To reiterate those findings, we would like to briefly touch on optimization in the early stages of planning chimeric molecules in a way that increases the likelihood of obtaining potent, cellpermeable degraders. In Figure 9, a representative ligand for each of the four most commonly used E3 ligases is presented along with its molecular descriptors. This radar plot can serve as a quick navigational tool to evaluate how the chosen E3 ligase ligand might contribute to the physicochemical properties of final degraders and aid in ligand selection in order to balance out the size, lipophilicity, and related characteristics that affect the success rate of PROTACs. Additionally, a selection of commercially available building blocks for the most widely applied CRBN, VHL, and IAP ligands are collected in Figure 10 and can be used as a quick aid for those starting with E3 ligase ligand synthesis.
This review gives an extensive overview of successful synthetic routes towards functionalized E3 ligase ligands. This enables the reader to better assess which reaction conditions are suitable and which yields can be achieved. Most of the starting materials are either commercially available or can be produced by simple synthesis techniques. Due to the scope of our research, this review may give new impulses in the synthesis laboratories to try out new linker connections or to test novel reactions under proven conditions. Ultimately, it is not only the accessibility and capital efficiency that determine the success of a PROTAC program, but also aspects such as rigidity, hydrolytic and metabolic stability, solubility and cell permeability of the chimeric molecules. This work represents a unique compendium to re-evaluate the many facets involved in the synthesis of such complex molecules.

AUTHOR CONTRIBUTIONS
AB, CS, RK, MG, and IS analyzed data and wrote the manuscript. MG and IS supervised the work.