Target validation is key to the development of protein degrading molecules such as proteolysis-targeting chimeras (PROTACs) to identify cellular proteins amenable for induced degradation by the ubiquitin-proteasome system (UPS). Previously the HaloPROTAC system was developed to screen targets of PROTACs by linking the chlorohexyl group with the ligands of E3 ubiquitin ligases VHL and cIAP1 to recruit target proteins fused to the HaloTag for E3-catalyzed ubiquitination. Reported here are HaloPROTACs that engage the cereblon (CRBN) E3 to ubiquitinate and degrade HaloTagged proteins. A focused library of CRBN-pairing HaloPROTACs was synthesized and screened to identify efficient degraders of EGFP-HaloTag fusion with higher activities than VHL-engaging HaloPROTACs at sub-micromolar concentrations of the compound. The CRBNengaging HaloPROTACs broadens the scope of the E3 ubiquitin ligases that can be utilized to screen suitable targets for induced protein degradation in the cell.

The development of proteolysis-targeting chimeras (PROTAC) confers a new modality of drug-like small molecules that have unique activities in inducing protein degradation by the ubiquitin-proteasome system (UPS) in the cell. [1] A typical PROTAC follows a heterobifunctional design in which ligands specific for an E3 ubiquitin ligase and a target protein are tethered through linkers of variable lengths to recruit target proteins to the E3 for proximity-based ubiquitination to signal their degradation by the proteasome. [2] While ligands of E3s have been developed to assemble PROTACs partnering with various E3s such as Cullin 2 (Cul2)-von Hippel-Lindau protein (VHL), Cullin 4 (Cul4)-cereblon (CRBN), and cellular inhibitor of apoptosis protein 1 (cIAP1), [1a,b,3] much effort is devoted to screening ligands against cellular targets to enable their degradation by the PROTACs. Before the commitment to the ligand discovery effort for specific cellular targets, it is desirable to screen a panel of cellular proteins to identify the ones that can be removed by UPS with high efficiency and, upon their removal, can generate the desired cellular response for therapeutical exploration. Methods for target identification and validation have been developed to tag the proteins of interest with a genetically encoded tag so that PROTACs functionalized with a tag-specific ligand can be used to recruit the target proteins to the E3 for ubiquitination, and subsequently, the dynamics of the target degradation and cellular response can be evaluated. So far, a variety of tags have been developed to assay protein degradation, including HaloTag, [4] a repurposed dehalogenase, the degradation tag (dTAG) based on variants of FK506-binding protein 12 (FKBP12), [5] BromoTag originated from the BRD4 bromodomain, [6] and the NanoLuc tag, an engineered luciferase. [7] HaloTag is a genetically modified bacterial dehalogenase that forms covalent bonds with ligands functionalized with terminal chlorohexyl moieties. [8] It is a popular choice for tagging cellular targets to validate their degradation by the UPS because the chlorohexyl group can be readily attached to E3 ligands through a linker to generate the bivalent HaloPROTAC that can trigger the ubiquitination and degradation of the tagged proteins (Figure 1). So far, HaloPROTACs engaging VHL and cIAP E3s have been developed, while the HaloPROTAC recruiting CRBN for induced protein degradation is missing. [4a,d,9] As most currently developed PROTACs utilize the VHL and CRBN ubiquitin ligases, we reasoned that the development of CRBN-recruiting HaloPROTACs would expand the versatility of this class of chemical probes as it has been shown that different E3 ligase ligands produce distinct degradation profiles. [5a] Furthermore, various cell and tissue types have substantially different expression levels of the E3s. [10] Thus, diversifying the E3-HaloPROTAC pairs could broaden the use of HaloTag-based platform for target validation in a larger bandwidth of cell types. Also advantageous in our minds, ligands that target CRBN (e. g., pomalidomide, thalidomide) are structurally simpler than the VHL and cIAP1 ligands, and thus chorohexyl-functionalized CRBN ligands present the opportunity to develop HaloPROTACs with streamlined synthetic routes and lower molecular weights to improve their pharmaceutical properties. Consequently, we set out to develop CRBN-recruiting HaloPRO-TACs to diversify the E3 pool engaged by HaloPROTACs in order to broaden the scope of the HaloTag-based target validation method. Herein we report the development of HaloPROTACs capable of recruiting the CRBN E3 ligase to induce protein degradation.

For our CRBN-recruiting HaloPROTAC design, we attached linkers of variable lengths with a terminal chlorohexyl moiety to the CRBN ligand pomalidomide at alternative positions and screened the activities of the HaloPROTAC molecules in inducing the degradation of EGFP fused to HaloTag2. Previously the EGFP-Halotag2 fusion has been used to validate the activity of hydrophobic tags for induced protein degradation. [4c,11] The reported design of VHL-recruiting HaloPROTAC-3 and HaloPRO-TAC-E demonstrated that degradation efficiency depended both on the point of linker attachment and linker length. [4a,d] Accordingly, in designing CRBN-recruiting HaloPROTACs, we attached linkers at either the 4-or 5-positions of the CRBN ligand as shown in Scheme 1. We also followed the Crews strategy of synthesizing longer linkers (15-21 atoms in length) since shorter linkers (6-12 atoms) were ineffective for VHLpairing HaloPROTACs. [4d] The synthesis of our CRBN-recruiting HaloPROTACs (Scheme 1) began with the addition of 1-chloro-6-iodohexane to ethylene glycols of various lengths to produce 1 a-1 c. These were then reacted with tert-butyl bromoacetate to construct tert-butyl ester-containing linkers 2 a-2 c. Ester hydrolysis using formic acid, followed by acid chloride formation using oxalyl chloride and catalytic DMF, produced acid chlorides 3 a-3 c, which were immediately coupled with commercially available pomalidomide (4-amino substitution) or the 5-amino substituted analog prepared via Brown's method. [12] This synthetic sequence rapidly produced CRBN-recruiting HaloPROTACs 4 a-4 c and 5 a-5 c with the linker attached to the 4-or 5-position of CRBN in yields of 24-63 % over the three-step sequence. The linker length in these HaloPROTACs varies between 15 and 21 atoms between the chlorine atom and the nitrogen attachment point to the thalidomide core. Using similar chemistry, we also prepared HaloPROTACs 4 d and 5 d with a 19-atom linker shown in Scheme 1 using the commonly employed 6-2-2-6 PROTAC linker developed by Crews. [13] As shown in Scheme S1, these HaloPROTACs were prepared in yields of 68 % and 52 % over a three-step sequence starting from the 6-2-2-6 linker containing the tert-butyl ester.

With our targeted library of eight novel CRBN-pairing HaloPROTACs in hand, we studied their ability to degrade HaloTagged EGFP in HEK293 cells following a previous report Cul2) as a scaffold that binds to the VHL through the bridging of Elongin B and C at one end and E2 through the bridging of Rbx1 at the other end. PROTAC ligands pairing with VHL such as HaloPROTAC-3 and HaloPROTAC-E would bind to VHL and covalently react with the HaloTag fused to the target proteins to recruit the targets to the E3 complex to induce their ubiquitination by proximity. [1] Polyubiquitinated target proteins would then be recognized by proteasome for degradation. This study developed a HaloPROTAC that is equipped with a CRBN ligand to recruit HaloTagged target proteins to the CRBN E3 complex consisting of Cullin 4 (Cul4), DDB1, and CRBN to induce their ubiquitination and degradation.

Scheme 1. Synthesis of CRBN-engaging HaloPROTACs.

on the development of HaloPROTAC-3. [4d] We first transfected the cells with the EGFP-HaloTag2 expression plasmid to induce its expression and then treated the cells with varying concentrations of the HaloPROTAC compounds ranging from 0.001 to 2 μM. After overnight incubation of the compounds with the cells, we measured the extent of EGFP degradation by fluorescence imaging of the cells and by western blotting of the cell lysates probed with an anti-EGFP antibody.

As 100 nM of 4 d) suggests that the increased aliphatic character of the linker affects the degradation efficiency. In addition to western blotting, we also imaged the fluorescence from cells expressing EGFP and found fewer cells with intense EGFP fluorescence after treating the cells with increasing concentration of HaloPROTACs of this series (Figure S2A). These results corroborate the western blot assays demonstrating the activity of CRBN-engaging HaloPROTACs in inducing the degradation of Halotagged EGFP as the target protein.

For HaloPROTACs with the linker at the 5-position, 5 c with a 21-atom linker is the most potent as it achieves more than 80 % degradation of EFGP-HaloTag2 at 1 μM and has an estimated DC 50 value of 2.8 nM (Figure S1B). In contrast, 5 a and 5 b containing shorter linkers were less active, yielding around 60 % degradation of EFGP-HaloTag2 at 1 μM of the compounds (Figure 2B). Interestingly, 5 d containing the 19-atom 6-2-2-6 linker structure displayed increased degradation over 4 d as it degraded ~80 % of EFGP-HaloTag2 at 1 μM, demonstrating that the 5-postion attachment point tolerates increased aliphatic character in the linker. Cell imaging studies for the 5-position series also showed that increasing the concentration of HaloPROTACs with a 5-positioned linker led to a lower number of cells with EGFP fluorescence (Figure S2B).

When comparing the 4-and 5-postion series, the best compound in the 4-series, 4 b, was more active than the best compound in the 5-series, 5 c. This was especially apparent when low compound concentration was used to induce protein degradation -at 1 nM, 4 b achieved more than 70 % degradation of EGFP-HaloTag2, while 5 c achieved a degradation of less than 30 %. Importantly, we also found both series of CRBNrecruiting compounds were tolerated by the HEK293 cells as no significant cell death was observed in an MTT assay after treating HEK293 cells overnight with concentrations of 4 b and 5 c up to 10 μM (Figure S3). This demonstrated that the protein degradation induced by treatment with HaloPROTACs 4 b and 5 c at increasing concentrations was not due to these compounds inducing cell death.

For comparing the CRBN and VHL-engaging HaloPROTACs in parallel, we synthesized HaloPROTAC-3 (Figures 1 and 3) that was previously developed as the degraders of HaloTagged proteins by pairing with the VHL E3. [4d] We found HaloPROTAC-3 can efficiently degrade EGFP-HaloTag2 fusion with a DC 50 value of 3.4 nM. The original report measured HaloPROTAC-3 activity in inducing the degradation of EGFP-HaloTag7 and reported a DC 50 of 19 nM. [4d] By comparison, CRBN-pairing degrader 4 b from this study matches the efficiency of HaloPROTAC-3, and since 4 b has a DC 50 value of 0.46 nM, it is more effective in inducing the degradation of EGFP-HaloTag2 at low compound concentrations. After treating the cell with 10 nM of the compounds, we found 4 b induced close to 80 % degradation of EGFP fusion while HaloPROTAC-3 induced about 60 % degradation (Figure 3).

To demonstrate the activity of 4 b in inducing the degradation of mammalian proteins, we constructed an expression plasmid of Aurora B kinase with C-terminal tags of EGFP-HaloTag2 and expressed the fusion protein in HEK293 cells. We found 4 b could degrade 50 % of the Aurora B fusion at a concentration as low as 10 nM and that ~80 % degradation was achieved with a concentration of 1 μM (Figure 2C without EGFP in HEK293 cells and treated the cells with varying concentrations of 4 b. We found 4 b can degrade Aurora B-HaloTag2 fusion with a similar efficiency for degrading Aurora B-EGFP-HaloTag2 fusion at concentrations greater than 1 μM (Figure 2C).

In conclusion, in this study, we developed HaloPROTACs that can utilize the CRBN E3 for degrading HaloTagged proteins and demonstrate the matching efficiency of the CRBN-recruiting ligands with the VHL-recruiting ligands in inducing degradation of HaloTagged EGFP as a model target. By varying the linker length and position of attachment to the pomalidomide core, we identified compound 4 b equipped with an 18-atom/3-PEG unit linker anchored at the 4-position of pomalidomide as the most efficient degrader (DC 50 = 0.46 nM), and it exceeds the efficiency of VHL-pairing HaloPROTAC-3 in inducing the degradation of HaloTagged EGFP at sub-micromolar concentrations of the compounds. We expect the CRBN-pairing 4 b can be broadly applied to target validation for PROTACS similarly to its VHL-pairing counterparts.

Besides the HaloPROTAC system, a variety of platforms for validating PROTAC targets have been developed, with the HaloPROTAC and dTAG systems being most broadly used. The lack of a CRBN compatible HaloPROTAC stands in contrast to the dTAG system that has ligands to promote degradation through both VHL and CRBN. [5] The development of CRBNengaging HaloPROTACs in this study provides a useful complement to the VHL-engaging HaloPROTACs and would expand the target validation platform for PROTAC development based on HaloTag fusions. The strength of the HaloTag platform stems from its generalizability to a variety of applications, including protein isolation and purification and the study of protein localization, function, and interactions. The commercial availability of 20,000 HaloTag-protein fusion proteins [4d] and the demonstrated ability to use CRISPR/Cas9 genome editing to generate HaloTag fusion proteins for endogenous proteins [4a,14] makes the HaloPROTAC platform applicable for investigating a wide range of potential therapeutic targets. The further extension of this readily available platform to targeted protein degradation amplifies the value of this technology as the installation of the HaloTag can be leveraged consecutively to study protein localization, interactions, and degradation. [15] The discovery of HaloPROTACs that engage the CRBN E3 ligase extends the versatility of the HaloTag/HaloPROTAC system for target exploration and validation through ubiquitination pathways mediated by various E3s.

Details of the experimental procedures are provided in the Supporting Information.