The rise of covalent proteolysis targeting chimeras

Abstract

Targeted protein degradation offers several advantages over direct inhibition of protein activity and is gaining increasing interest in chemical biology and drug discovery. Proteolysis targeting chimeras (PROTACs) in particular are enjoying widespread application. However, PROTACs, which recruit an E3 ligase for degradation of a target protein, still suffer from certain challenges. These include a limited selection for E3 ligases on the one hand and the requirement for potent target binding on the other hand. Both issues restrict the target scope available for PROTACs. Degraders that covalently engage the target protein or the E3 ligase can potentially expand the pool of both targets and E3 ligases. Moreover, they may offer additional advantages by improving the kinetics of ternary complex formation or by endowing additional selectivity to the degrader. Here, we review the recent progress in the emerging field of covalent PROTACs.

Keywords : Covalent PROTACs, Targeted degradation, Chemoproteomics, E3 li- gases, Reversible covalent.

Introduction

The concept of targeted protein degradation (TPD) has emerged in recent years as a powerful approach to modulate protein activity for therapeutic or chemical biology applications. The most common strategy to achieve TPD is proteolysis targeting chimeras (PROTACs) [1,2]. These heterobifunctional molecules comprise an E3 ligaseebinding moiety tethered to a target-binding group through a linker (Figure 1). By inducing a ternary complex between the target and E3 ligase, PROTACs facilitate target ubiquitination and subsequent degradation. Other emerging methods to achieve TPD include molecular glues [3e6], autophagy-mediated degraders [7,8], and additional methods that expand the range of targets to include extracellular proteins [9]. TPD offers several benefits over direct inhibition of protein targets, such as improved pharmacodynamics due to the need for resynthesis of the protein target [10], the abolishment of all functions and interactions of the target due to its degradation, as well as improved selectivity over the target binder alone [11,12].

In order for PROTACs to induce degradation of their target protein, the degrader molecule must first bind both the target and the E3 ligase. Developing molecules with sufficiently high binding affinity to the target and identifying new potent E3 ligase binders are among the main barriers to the development of potent degraders because many targets have proven recalcitrant to ligand discovery [13e15], and efficient recruiters are so far popular for only a handful of E3 ligases, such as cereblon [16], VHL [17], IAP [18], and MDM2 [19].

One powerful approach to overcome this problem is the use of covalent binding to the target and/or E3 ligase. Covalent inhibitors are powerful therapeutic and chemical biology tools owing to their high potency, su- perior pharmacodynamics, and the potential for improved selectivity. An exceptional example is KRAS, which is frequently mutated in cancer and was long considered an ‘undruggable’ target. Covalent targeting of the G12C mutation in KRAS [20] unlocked it as a target for therapy, and as of date, three covalent in- hibitors of KRASG12C have entered clinical trials [21e 24].

Indeed, the first PROTAC reported by Sakamoto et al. [25], which targeted methionine aminopeptidase-2, used the covalent inhibitor ovalicin as the target bind- ing moiety. However, the vast majority of PROTACs bind noncovalently to their target (Figure 1). It was postulated that covalent binding would not enable the PROTAC molecule to engage multiple target molecules (Figure 2a), thus negating the catalytic nature of degradation by PROTACs [26]. In a prominent example, Tinworth et al. [27] prepared acrylamide PROTACs targeting Bruton’s tyrosine kinase (BTK). While the acrylamide version of the PROTAC,degraders are being constantly developed and offer several benefits over noncovalent degraders. Here, we review the recent developments in this field.

PROTACs that engage the target via an irreversible covalent bond. (a) Outline of the degradation mechanism involving irreversible covalent engagement.(b) Published examples of irreversible covalent PROTACs. The names given to the molecules are taken from the publications in which they are described. The electrophile is colored in red, the target reversible recognition group is colored in blue, the linker is colored in black, and the E3 ligase binder is colored in violet. PROTAC, proteolysis targeting chimera.

Covalent binding to the degradation target Although covalent binding to the target protein (Figure 2a) could theoretically negate the catalytic behavior of a PROTAC, it can still be highly bene- ficial for PROTAC development. In the most obvious case, covalent binding enables the degradation of targets for which potent noncovalent binders are not available owing to the lack of well-defined pockets or very high affinity to natural substrates, as in the case of KRAS. Zeng et al. [28] aimed at developing KRASG12C degraders based on the covalent inhibitor ARS1620 [29]. They developed a high-throughput cell sorting assay that monitors the fluorescence of a GFP-KRAS fusion protein, enabling the rapid quantitative screening of different linker chemistries. While the screen led to the discovery of potent de- graders of the GFP fusion protein, they failed to degrade endogenous KRASG12C. In a later work, Bond et al. [30] synthesized PROTACs derived from MRTX849 [22] that degrade KRASG12C potently in a proteasome-mediated mechanism. In a second example, Buckley et al. [31] developed the concept of HaloPROTACs, that use a HaloTag-binding group to degrade HaloTag-labeled proteins. HaloPROTACs have proven to be versatile and flexible chemical biology tools for rapid, chemical knockdown of target proteins [32,33].

However, PROTACs that engage their target covalently have displayed promising results even for targets where potent reversible binders exist. In one example, Lebraud et al. [34] designed ERK1/2-targeting PROTACs that form in situ by cycloaddition of tetrazine- labeled thalidomide and a trans-cycloocteneelabeled ERK1/2 covalent inhibitor. The authors observed potent degradation of ERK1/2, which was partly mediated by noncovalent binding and partly by covalent binding. Burslem et al. [35] observed efficient degradation of a gefitinib-resistant mutant of Epidermal Growth Factor Receptor (EGFR) by a PROTAC using a covalent, afatinib-based warhead. Recently, Xue et al. [36] developed acrylamide-based PROTACs that degrade BTK and the off-target BLK. These covalent PROTACs exhibited equal or superior degradation potency relative to their noncovalent analogs (although not nearly as potent as other known BTK degraders) and engaged BTK covalently in the cell.

A possible approach to solve the problem of noncatalytic degradation by covalent PROTACs is the utilization of ‘reversible’ covalent chemistry to mediate the binding to the target (Figure 3). In theory, such PROTACs would exhibit improved target binding and support catalytic degradation. Several such examples have very recently appeared. de Wispelaere et al. [37] developed degraders for hepatitis C virus NS3 protease based on telaprevir, which forms a reversible hemiketal with the active-site serine. One of the degraders, DGY-08-097, potently inhibited viral replication primarily by degrading the protease and not by direct inhibition and also inhibited viral variants that were resistant to telaprevir. This result exemplifies the potential of event-driven inhibition in tackling resistant viral vari- ants because sustained high occupancy of the target is not necessary to achieve extensive degradation. In a second example, Peng et al. [38] developed highly potent (w100% degradation at <100 nM) cyanoacrylamide-based degraders of estrogen-related receptor alpha, which plays important roles in meta- bolic regulation. However, the role of covalent binding in this case is unclear, as a covalent interaction with an exposed cysteine was only predicted by computational studies, and the b and g isoforms, which harbor the same cysteine, are not degraded.

Examples of reversible covalent PROTACs. The names given to the molecules are taken from the publications in which they are described. The elec- trophile is colored in red, the target reversible recognition group is colored in blue, the linker is colored in black, and the E3 ligase binder is colored in violet. ERRa, estrogen-related receptor alpha; PROTAC, proteolysis targeting chimera.

Two particularly interesting and well-characterized ex- amples of reversible covalent PROTACs derived from ibrutinib and targeted against BTK have been published by Guo et al. [39] and by our laboratory [40]. Guo et al. [39] synthesized a cyanoacrylamide-based PROTAC (RC-1) that exhibited highly potent BTK degradation in cell culture and in mice and had favorable pharmaco- logical properties. RC-1 exhibited superior degradation and higher selectivity than its noncovalent and irre- versible covalent (acrylamide) counterparts. Estimation of BTK and Cereblon engagement using NanoBRET [41] indicated that the improved potency of RC-1 was mainly due to its higher cellular permeability. However, RC-1 degraded the C481S BTK mutant with similar potency and also potently degraded CSK, a noncovalent off-target of ibrutinib. These results may indicate that RC-1 degrades BTK primarily via noncovalent binding
in cells. Using similar chemistry, Guo et al. [39] pre- pared cyanoacrylamide-based PROTACs against fms- like tyrosine kinase 3 (RC-FLT3; Figure 3).

In a separate and parallel study, we developed nonco- valent (NC-1), cyanoacrylamide-based (RC-3), and acrylamide-based (IR-2) PROTACs, which all degraded BTK potently with DC50 <10 nM and had similar cellular permeabilities. Similar to RC-1 in the study by Guo et al. [39], acrylamide IR-2 degraded and inhibited BTK C481S with similar potency as the wild-type one and degraded noncovalent off-targets of BTK such as CSK and LYN, indicating that it drives degradation primarily through noncovalent binding. In contrast, the cyanoacrylamide RC-3 displayed much higher potency
toward wild type (WT) BTK than the cysteine mutant and did not degrade noncovalent off-targets, indicating that it recruits BTK for degradation exclusively through covalent binding, which led to enhanced selectivity. However, we discovered that although RC-3 bound BTK reversibly, its dissociation rate was very slow relative to the degradation rate, indicating that it may only weakly support a catalytic degradation mechanism.

Therefore, understanding how covalent binding affects degradation by each PROTAC requires careful measurement of the rate of covalent bond formation and dissociation. Importantly, both Guo et al. [39] and our group devel- oped covalent BTK PROTACs with nanomolar po- tencies, which is comparable with the published highly potent BTK PROTACs based on reversible binders [42] (Table 1).

PROTACs is not straightforward as the degree of true covalent engagement to the target is not always known, especially if the ligand has high affinity in the absence of covalent binding. Nevertheless, although covalent PROTACs can occasionally compete and even surpass noncovalent PROTACs in their potency, we believe their greatest potential still lies in the ability to degrade targets without potent reversible binders and their po- tential for enhanced selectivity.

Covalent binding to the E3 ligase

So far, only a handful of E3 ligases have been consis- tently used for targeted degradation, all with noncova- lent ligands. Because there are hundreds of E3 ligases in the proteome [43], it is postulated that developing recruiting ligands for new ligases will vastly increase the repertoire of degradable targets. In this discovery pro- cess, covalent chemistry is particularly useful d cova- lent binders can be screened rapidly using mass spectrometry [44e48], and new targets can be identi- fied using proteomics methods such as isoTOP-ABPP [49e53]. Covalent binding to the E3 ligase can also offer some intrinsic advantages over noncovalent bind- ing d first, covalent binding may enable targeting of sites that are difficult to bind with noncovalent re- cruiters, especially sites distinct from the active site, thus enabling efficient recruitment of the E3 ligase without compromising its activity. In addition, an irre- versible E3ePROTAC complex can recruit multiple target molecules for ubiquitination and degradation, without the need for the prior kinetic step of forming the E3ePROTAC interaction. This could significantly improve the kinetics of ternary complex formation (Figure 4a). Several PROTACs targeting novel E3 li- gases via this mechanism have been developed, although they still require optimization to improve their potency and selectivity (Figure 4b).

PROTACs that engage the E3 ligase through a covalent bond. (a) Model for improved degradation kinetics due to covalent recruitment of E3 ligases. The binary complex between the PROTAC and the E3 ligase is formed irreversibly and does not need to reform between degradation cycles. (b) Published examples of novel covalent E3 ligase recruiters. The names given to the molecules are taken from the publications in which they are described. The electrophile is colored in red. PROTAC, proteolysis targeting chimera.

Ward et al. [54] sought to find binders for the E3 ligase RNF4, which targets SUMOylated proteins for ubiquitination and degradation. Gel-based activitye based screening of electrophilic compounds followed by optimization of the scaffold led to CCW16, which binds a cysteine in the zinc finger of RNF4 while not interfering with zinc binding by the neighboring cys- teines, thus maintaining the protein folded and active. A PROTAC containing CCW16 attached through a linker to the bromodomain inhibitor JQ1 degraded BRD4 selectively and potently without degrading BRD2 and BRD3.

In a second example, Zhang et al. [55] used an opposing approach d they attached broadly reactive electrophilic fragments to a binder of the prolyl isomerase FKBP12, and after identifying a fragment that led to FKBP12 degradation, they used pull- downebased proteomics to identify the E3 ligase bound by the fragment, leading to the identification of DCAF16. Interestingly, only a fraction of DCAF16 had to be labeled by the PROTAC to drive degrada- tion, indicating that most of DCAF16 can still perform its regular function.

Another E3 ligase that was explored covalently for targeted degradation is RNF114. isoTOP-ABPP revealed that the natural product nimbolide binds to cysteine 8 in RNF114, which inhibits p21 degradation and induces apoptosis. A PROTAC derived from nimbolide and the bromodomain inhibitor showed potent BRD4 degradation [56]. Later, Luo et al. [57] used gel-based ABPP with purified RNF114 to identify fragment-sized covalent binders of RNF114. One of the hits, EN219, showed moderate selectivity, did not label other E3 ligases, and was used to prepare potent PROTACs targeted against BRD4 and BCR-Abl.

In another seminal study, Tong et al. [58] developed BRD4-degrading PROTACs based on the KEAP1-Nrf2 activator bardoxolone, which forms reversible covalent interactions with KEAP1 via a cyanoenone moiety. The bardoxolone-based PROTAC potently degraded BRD4, and furthermore, the authors showed that covalent binding to KEAP1 was essential for degradation as reduction of the double bond or attenuation of elec- trophile reactivity by removal of the cyano group abol- ished the degradation activity.
Table 2 summarizes the data regarding the PROTACs that covalently recruit novel E3 ligases. Some of these PROTACs exhibit potencies similar to known PROTACs that recruit established E3 ligases such as CRBN and VHL.

More covalent binders or inhibitors for several other E3 ligases, such as KEAP1 [59], HOIP [60], and Nedd4 [61], have been found, but their potential for targeted degradation remains to be explored. To estimate the future potential of covalent recruitment of E3 ligases for targeted degradation, we curated the data from several chemoproteomics studies that identified ligandable cysteines in the proteome using various chemical probes [52,53,62e66] and cross- referenced them to a list of E3 ligases and other associated proteins that can serve for recruitment [67]. We identified ligandable cysteines in 78 pro- teins, including proteins already successfully used for targeted degradation such as RNF114 and KEAP1 (Table 3). Although the E3 covalent binders discovered so far still need to be optimized to improve their binding and selectivity, we believe they will mature into powerful tools for protein degradation.

Outlook and conclusion

Although most of the covalent degraders we described so far are PROTACs, the concepts described here could also be applied to others types of degraders such as molecular glues. In one example, Isobe et al. [6] char- acterized the polyketide asukamycin, which covalently recruits the E3 ligase UBR7 and forms a ternary complex with p53, which in this case resulted in stabilization rather than ubiquitination of p53. However, this implies the potential to use covalent molecular glues to recruit E3 ligases to target proteins. The advantages of covalent binding are expected to extend to numerous other modalities.

In conclusion, the use of covalent binders can offer several benefits for targeted degradation: it can expand the repertoire of possible targets and enhance the po- tency or selectivity of the degraders. It may also facili- tate identification of new protein components to facilitate new modalities of degradation. We therefore expect that covalent chemistry and covalent ligand discovery will play an important role in future development of YD23 targeted degraders.