Identification of ligand linkage vectors for the development of p300/CBP degraders

Abstract

To develop new degrader molecules from an existing protein ligand a linkage vector must be identified and then joined with a suitable E3 ligase without disrupting binding to the respective targets. This is typically achieved through empirically evaluating the degradation efficacy of a series of synthetic degraders. Our strategy for determining optimal linkage sites utilises biotinylated protein ligands, linked via potential conjugation sites of an inhibitor to confirm whether target protein is maintained after forming a conjugate. This method provides low-cost, qualitative evidence that the addition of a linker moiety at a specific position can be tolerated, guiding further optimisation. We demonstrate the application of this method through the exploration of linkage vectors on A-485, a known ligand of p300/CBP, and found a conjugation site through a urea moiety. Pomalidomide was then conjugated through this site with several different linkers and cell viability and degradation were assessed for this library using a myeloma cell line, MM1.S. Compound 18i, with a PEG₄ linker, was found to be the most effective p300 degrader and linker length greater than 10 atoms afforded enhanced degradation.

To develop new degrader molecules from an existing protein ligand a linkage vector must be identified and then joined with a suitable E3 ligase without disrupting binding to the respective targets.

Introduction

Targeted protein degradation is a rapdily expanding field in chemical biology and medicinal chemistry that relies on the repurposing of endogenous degradation pathways towards a neosubstrate, in order to develop novel therapeutics and chemical probes. Examples of these degrader molcules include molecular glues, such as thalidomide and its related analogues,^1–3 and proteoylsis targeting chimeras (PROTACs) that redirect the ubiquitin proteosome system to degrade specific targets.^4–8

A demonstrated strategy for the successful design of heterobifunctional degraders is through the repurposing of inhibitors which have suboptimal drug-like properties, but exhibit a high degree of specificity.⁹ Even with an inhibitor in hand, the site for conjugation to an appropriate E3 ligase ligand is frequently not obvious as inappropriate sites of conjugation can disrupt binding to the target protein.¹⁰ In addition, orientation of the target protein and E3 ligase can vary with different linkage vectors, altering the efficacy and specificity of degradation.^11–13 Different orientations of both proteins will in turn affect cooperativity for the formation of the requisite ternary complex.^14,15

After obtaining all available structural information, a low-cost strategy that has been employed in our lab is to synthetically introduce a linker and biotin tag at candidate sites on the lead targeting ligand. These biotin probes can then be incubated with cellular lysate, followed by routine immunoprecipitation and western blot analysis. Visualisation of the target protein can act as a qualitative identification for a site of conjugation, often referred to as the linkage vector of a heterobinfunctional degrader.¹⁰ Alternative approaches may utilise computational modeling to predict conjugation sites,^16–19 or empirically synthesising and testing various different linkage vectors by determining degradation or binding affinities to the targeted protien. While these strategies have also proven effective, the use of biotin probes can rapidly determine viable linkage vectors to act as an initial checkpoint in the development of new heterobifunctional degraders, followed by empirical optimisation and tuning of linker properties.

The paralogous proteins p300 and CBP are epigenetic enzymes that play an important role in gene expression through acetylation of histones, transcription factors and other proteins.^20,21 While p300/CBP also have roles in cell proliferation, maintenance, tumor supression and many additional functions, the misregulation of these enzymes is commonly associated with several cancers, including multiple myeloma.^22,23 Therefore, targeting p300/CBP, specifically their histone acetyltransferase (HAT) domains, has become of interest.^22,24–31 Our group planned to target p300/CBP using a protein degradation strategy as they would be expected to be useful as potential therapeutics and chemical probes to study biological interactions through knockdown experiments.^32–34 Recent elegant examples from the Ott group and the Qi group have highlighted the utility and potential of p300/CBP degraders.^24,35 We aimed to demonstrate the utility of our biotin conjugation strategy to validate linkage vectors of p300/CBP inhibitors. The p300 HAT ligand A-485 was identified as a viable starting point and was conjugated with biotin in order to validate different sites as possible linkage vectors (Fig. 1).^22,25

Fig. 1. Structures of A-485 and analysis of X-ray structure of A-485 bound to p300 (PDB: 5KJ2) A) structure of A-485 with potential conjugation sites. B) Solvent exposed methyl urea of A-485 bound to p300. C) Potential conjugation site through homologation of methyl group.

Results and discussion

Identification and validation of A-485 linkage vectors

Analysis of the X-ray structure of A-485 bound to p300 HAT (PDB: 5KJ2) revealed two potential linkage vectors (Fig. 1). The methyl group attached to the urea (Fig. 1a) and the chiral trifluoromethylated N-alkyl substituent (Fig. 1b) of A-485 both appear to be solvent exposed and could be functionalised for conjugation. With potential linkage vectors identified, we embarked upon the synthesis of biotinylated probes of A-485, beginning with a urea-linked biotinylated probe 2 (Scheme 1). The penultimate A-485 precursor, aromatic amine 1, was prepared following the procedure established by AbbVie,^22,25 which in turn was converted to an isocyanate using triphosgene. Biotinylated amine linker 13 was then added to the solution to afford 2 as a urea linked A-485 biotin probe, which was then subjected to incubation with solid-phase bound streptavidin in cell lysate. Following the pulldown experiment and western blot analysis for p300, we observed that biotin probe 2 was able to bind to the desired target (Fig. S1†). These results suggested that conjugation through the urea linkage is compatible with p300 binding, providing support for synthesis of urea-linked degraders.

Scheme 1. Preparation of biotin conjugate 2.

Our focus turned towards the synthesis of a trifluoromethylated N-alkyl derivative required for the installation of a linker at this second linkage vector (Fig. 1b, Scheme 2). Utilising Ellman's sulfinamide as a chiral auxiliary, a trifluoromethyl group was added to the sulfinamide via condensation to give intermediate 3, followed by addition of homoallyl Grignard reagent to afford 4 as a single diastereomer, whose absolute configuration was confirmed by single crystal X-ray diffraction (Scheme 2).

Scheme 2. Synthesis of A-485–biotin conjugate 14.

Sequential N-benzylation, removal of the chiral sulfinamide and amidation provided the bromo acetamide 7 in 67% yield over three steps. Coupling of 7 with the enantiopure spirocycle 8 (Scheme S1†),²⁵ gave 9, which upon Buchwald–Hartwig amination with diphenylimine followed by acid hydrolysis gave aromatic amine 10. Next, 10 was converted to an isocyanate in situ using triphosgene followed by addition of methylamine to produce the A-485 homoallyl derivative 11. The key oxidative cleavage of the alkene of 11 to carboxylic acid 12 proceeded in modest yield to enable subsequent conjugation to N-(4-aminobutyl)biotinamide 13, providing A-485-derived biotin probe 14 through the second candidate conjugation site. Conjugate 14 was then incubated with OPM2 cell lysate with solid-phase bound streptavidin but p300 was not observed via pulldown and Western blot analysis, suggesting that this site was not amenable as a linkage vector for potential p300 degraders.

Synthesis of pomalidomde-based A-485 conjugates for p300 degradation

With evidence for a linkage vector from our biotin–streptavidin method, we next made conjugates of A-485 with an E3 ligase ligand, specifically pomalidomide due to its ease of synthesis and its effective use as a ligand for cereblon.^36–38 Pomalidomide-linkers with Boc protected amine handles, 16, were deprotected using TFA and the resulting trifluoracetate amine salts, 17, were then used for the next conjugation step (Scheme 3). These pomalidomide-linkers were conjugated to A-485 via triphosgene mediated urea formation from in situ generated isocyanate from 15. A-485–pomalidomide conjugates were incubated with MM1.S cells for 24 hours and p300 levels were assessed using western blot analysis (Fig. 2 and S5†). Compound 18i was found to be the most potent compound in the range tested from 1 to 10 μM. Maximal degradation was observed around 10 μM for 18i, with approximately 9% of p300 remaining. As well, this linker was also found to be effective for dCBP-1, a previously reported GNE-781 based CBP/p300 degrader, where the authors predicted a minimal distance of 20 Å for effective ubiquitination, the approximate length of a PEG₄ linker.²⁴ Linker length was also compared against p300 degradation (Fig. 2). Linkers of A-485–pomalidomide conjugates were clustered as either short (<9 atoms in length) or long (>10 atoms in length) compared to amount p300 remaining after treatment at 10 μM. Shorter linkers were on average found to be poorer p300 degraders than longer linkers with 42% and 18% p300 remaining, respectively, in agreement with previous studies.²⁴ Compounds were also evaluated for their impact on cell viability in MM1.S cells. Only a modest correlation with p300 degradation was observed, suggesting these compounds may undergo alternative mechanisms to inhibit cellular growth of MM1.S cells. Compounds 18i, 18e and 18h showed the best in vitro degradation of p300 despite exhibiting a modest impact on cell viability of MM1.S cells. Consistent with our finding that conjugation through the aryl amine position of A-485 is suitable for protein degradation and p300 engagement, a recently reported degrader JQAD1,³⁵ also utilized the same linkage vector as identified in this work. Based on the degradation results, conjugates 18i was incubated with MM1.S cells for 72 hours and inhibition of cell proliferation was assessed (Scheme 3, Fig. S2†). Compounds 18g and 18i were found to have a comparable impact on MM1.S cell viability as A-485 (294, 422 nM and 139 nM, respectively).

Scheme 3. Preparation of A-485–pomalidomide conjugates.

Fig. 2. Cellular degradation of p300 using A-485–pomalidomide conjugates at 10 μM concentration. A-485–pomalidomide conjugates were clustered based on linker length as either short (<9 atoms) or long (>10 atoms) linkers and were found to be significantly different (p = 0.018).

Conclusions

We have demonstrated the use of biotin conjugates to explore potential linkage vectors on protein inhibitors en route to developing new degraders. The use of existing crystallography data helped guide the placement and exploration of linkage vectors. The use of biotin conjugates with streptavidin magnetic beads followed by immunoprecipitation offers a fast route for identifying potential linkage vectors to guide PROTAC design. Using biotinylated probes, we examined two potential linkage vectors on the p300 HAT inhibitor A-485 and found that the urea functional group was amenable towards conjugation. We synthesized a library of nine A-485–pomalidomide conjugates that were tested for their effects on cellular viability and p300 degradation in a myeloma cell line, MM1.S. Compound 18i containing a PEG₄ linker, was found to be effective in the cell viability assay, and the most efficient p300 degrader of our candidate compounds.

Conflicts of interest

There are no conflicts to declare.