A Phenotypic Approach for the Identification of New Molecules for Targeted Protein Degradation Applications

Abstract

Targeted protein degradation is an emerging new strategy for the modulation of intracellular protein levels with applications in chemical biology and drug discovery. One approach to enable this strategy is to redirect the ubiquitin–proteasome system to mark and degrade target proteins of interest (POIs) through the use of proteolysis targeting chimeras (PROTACs). Although great progress has been made in enabling PROTACs as a platform, there are still a limited number of E3 ligases that have been employed for PROTAC design. Herein we report a novel phenotypic screening approach for the identification of E3 ligase binders. The key concept underlying this approach is the high-throughput modification of screening compounds with a chloroalkane moiety to generate HaloPROTACs in situ, which were then evaluated for their ability to degrade a GFP-HaloTag fusion protein in a cellular context. As proof of concept, we demonstrated that we could generate and detect functional HaloPROTACs in situ, using a validated Von Hippel–Lindau (VHL) binder that successfully degraded the GFP-HaloTag fusion protein in living cells. We then used this method to prepare and screen a library of approximately 2000 prospective E3 ligase-recruiting molecules.

Keywords

targeted protein degradation E3 ubiquitin ligases HaloTag PROTACs HTE phenotypic screening

Introduction

Protein degradation has emerged as an important therapeutic strategy in which small molecules can recruit cellular protein degradation machinery to induce the degradation of selected proteins of interest (POIs). One important subset of degraders that are targeted to specific POIs are proteolysis targeting chimeras (PROTACs), which are bifunctional molecules composed of a ligand to the POI linked to a second ligand that binds an E3 ligase.^1–6 Ternary complex formation between the target POI, the PROTAC, and the E3 ligase enables the POI to be covalently modified with polyubiquitin chains, which are then recognized by the proteasome.^7–10 Many different POIs have been reported to be successfully degraded by PROTACs, and several PROTAC molecules have now entered clinical trials.^11–13 Currently reported PROTACs recruit only a small proportion of the known E3 ligases encoded by the human genome.^14–19 However, expanding the available portfolio of functional E3 ligase-recruiting ligands might provide additional versatility for the degradation of challenging targets, as well as potentially offering improvements in physicochemical properties and developability as compared with currently known E3 ligase chemical binders.^20–22

One approach to expanding the chemical matter available for the recruitment of E3 ligases is to conduct high-throughput screening campaigns against specific individual E3 ligases using binding affinity methods such as encoded library technology.^23,24 This type of screening has the advantage of being high throughput and well established.^25–27 However, this approach also means that screened E3 ligases must generally be selected with limited knowledge about their substrate scope and tractability. Additionally, even before a functional PROTAC can be constructed from an E3 ligase screening hit, a medicinal chemistry effort is typically required to improve E3 ligase binding affinity and to prepare and test multiple putative PROTACs with various linkers and linker attachment configurations. In practice, this reductionist single-target screening approach for E3 ligases is costly, inefficient, and slow, and can realistically only be conducted on a small fraction of the total number of potential E3 ligases that might have utility in PROTAC design.²⁸

To address the inherent inefficiencies posed by serial screening of individual E3 ligases, we have developed and implemented an alternative strategy that combines a cell-based targeted protein degradation phenotypic screen with efficient 384-well high-throughput chemical synthesis, shown schematically in Figure 1 . We based our screening design strategy on the previously described HaloPROTAC method for the evaluation of Von Hippel–Lindau (VHL) E3 ligase binders in a cellular context. HaloPROTAC molecules contain a linker with a terminal chloroalkane group that can covalently bind to the HaloTag protein,²⁹ and we have previously demonstrated the degradation of a HaloTag-GFP (green fluorescent protein) fusion protein upon cell treatment with a VHL-based HaloPROTAC.³⁰ We reasoned that the HaloPROTAC assay could be developed into a high-throughput phenotypic screening format, where new ligands would be identified based on their ability to act as functional degraders when tethered to the HaloTag-GFP protein in a cell. This format should be agnostic regarding the underlying mechanism of POI degradation but would effectively multiplex a screening campaign of E3 ligases as well as other intracellular proteins and mechanisms relevant to targeted protein degradation. Any resulting hits will have already demonstrated functional target degradation in cells and would in principle be suitable for incorporation into heterobifunctional degraders.

Figure 1.

Schematic overview. (A) HaloPROTACs 3 are synthesized in situ in 384-well plates in high-throughput format using an optimized amide coupling reaction between a reactive haloalkane-containing linker 1 and a library of amines 2. (B) HaloPROTACs 3 are screened in a 384-well-format HeLa cell-based phenotypic assay for the ability to degrade HaloTag-GFP. This may be achieved by recruitment of an E3 ligase such as with a PROTAC, or via recruitment of other cellular degradation mechanisms. (C) False positives are identified in a counterscreen using a D106A HaloTag mutant that is incapable of binding the HaloPROTAC.

To realize the goal of conducting a high-throughput phenotypic HaloPROTAC screen, the preparation of a suitable HaloPROTAC screening library must be addressed, as we predicted that this step might be rate-limiting. We hypothesized that it might be feasible to prepare a HaloPROTAC library 3 in situ, via high-yielding coupling reactions in 384-well low-volume plates between a diverse array of amines 2 and an activated ester building block 1 containing a linker and a primary chloroalkane ( Fig. 1A ). This approach provides for significant diversity, since multiple HaloPROTAC screening libraries can be prepared from the same amine building blocks by employing different chloroalkane linkers.

Since only fully elaborated HaloPROTAC molecules should be capable of degrading the HaloTag-GFP fusion protein, we further hypothesized that it would be possible to screen the HaloPROTAC library in situ, without purification of the individual library compounds. We also speculated that in view of the multiplexed nature of this assay design, which in principle screens all of the E3 ligases and degradation mechanisms present in the cells simultaneously, it might be possible to identify a functional degrader hit from a much smaller compound library than would otherwise be required in an equivalent nonmultiplexed single-target screening campaign. Toward this goal, we describe herein the optimization of a high-throughput amide coupling reaction in low-volume 384-well format, and the efficient synthesis of a HaloPROTAC screening set starting from a curated library of approximately 3000 reactive amines. The resulting HaloPROTAC screening set was used to conduct a phenotypic screening campaign to detect functional HaloTag-GFP degraders in HeLa cells ( Fig. 1B ). A false-positive counterscreen was also implemented using a cell line with a mutated version of the HaloTag protein that does not covalently bind chloroalkanes ( Fig. 1C ). Our work demonstrates a novel application of HaloTag technology for a phenotypic screening approach to detect small-molecule ligands for use in targeted protein degradation.

Materials and Methods

General High-Throughput Chemistry Considerations

Experiments conducted under an inert atmosphere were performed inside a MBraun glovebox with a constant N₂-purge with oxygen typically <5 ppm. Stock solution reservoirs were polystyrene, 50 mL capacity (Corning Costar, New York, US, cat. 4870); 384-well plates were polypropylene and either Greiner v-bottom (cat. 781280) or Labcyte flat-bottom (cat. P-05525). Plates were covered with a polystyrene lid where indicated (Corning, cat. 3098). Pipetting procedures were completed using an electronic multichannel pipette from stock solution reservoirs (Thermo Fisher, 12 channels, 1–30 μL, cat. 4671030BT with Thermo Fisher ClipTip, cat. 94410103, Cheshire, UK).

HaloPROTAC Library Design and Synthesis

The HaloPROTAC screening library 5 was constructed via the high-throughput coupling of a diverse array of amines 2 with a single activated N-hydroxysuccinimide (NHS) ester³¹ building block 4 bearing a primary alkyl chloride ( Fig. 2A ). Suitable amines were selected based on their predicted reactivity under the anticipated reaction conditions, the presence of a chromophore to facilitate UV detection, structural diversity, and the predicted cellular permeability of the corresponding fully elaborated HaloPROTACs. Based on these criteria, a total of 2934 amines (661 primary, 2267 secondary, and 6 compounds containing both primary and secondary amines) were selected for use in library construction from the GlaxoSmithKline (GSK) compound collection. Supplemental Figure S1 provides an overview of the physicochemical property distribution of the selected amines, and details of the selection procedure are described in the Supplemental Material.

Figure 2.

(A) HaloPROTAC 5 synthesis from the selected amines 2 through an optimized amide coupling with an NHS ester 4. (B) Example LC-MS analysis of plate 2 of nine plates, showing reactions with >50% conversion in yellow, <20% conversion in blue, and in-between in green. (C) Overall outcome for the reactions, out of the 2934 amines. (D) Computational analysis of the reactions, showing successful conversion in yellow and categorized by amine steric hinderance.

HaloPROTAC library synthesis was conducted directly on 384-well reactor plates containing the selected amines in random order. The plates were prepared to contain 10 μL of 10 mM DMSO stock solution per well and were stored at −20 °C. The selected amine building blocks were delivered on nine plates (including 16 negative control DMSO wells and 16 wells containing the VHL ligand 6 corresponding to positive control VHL-HaloPROTAC 7) ( Fig. 3A ), which were also synthesized in situ (example plate layout shown in Fig. 2B ). The plates were thawed at room temperature, centrifuged briefly at 200g, and then placed in a glovebox prior to reaction dispensing. Reaction wells were then dispensed with stock solution 1 (SS1; prepared as described below), and the reactor plates were then sealed with lids (Greiner, cat. 676001) and left to stand at room temperature for 16 h. The lids were then removed, DMSO (20 µL) was added to each well, and each plate was removed from the glovebox and resealed using a new metal foil lid. The plates were then placed into an automatic sample organizer and analyzed by liquid chromatography−mass spectrometry (LC-MS) for reaction completion and the presence of the target mass in each well.

Figure 3.

(A) Structures of a reference VHL ligand 6 and two positive control VHL-HaloPROTACs 7 and 8, and validation of the HeLa HaloTag-GFP screening and HaloTag (D106A)-GFP counterscreening cell lines. (B) Example fluorescence images taken in 384-well format showing differential GFP degradation (fluorescence reduction) from 1 µM control VHL-HaloPROTAC 7 treatment. (C) Concentration−response profiles using different incubation times with fluorescence imaging. Plots use the average values of two technical well replicates. Cells in this experiment were preclonal selection, 20,000 cells per well.

For SS1, a stock solution of the chloroalkane NHS ester 2,5-dioxopyrrolidin-1-yl 18-chloro-3,6,9,12-tetraoxaoctadecanoate 4 (42.4 mg, 0.100 mmol) ( Fig. 2A ) and N-methylmorpholine (NMM; 109 µL, 0.991 µmol) in DMSO (10 mL) was prepared in a glovebox under a nitrogen atmosphere to give a 10 mM DMSO solution that was used immediately.

An analysis of the reaction outcomes for each of the nine plates is shown in Supplemental Figure S4 .

Stable Cell Line Preparation

A HeLa cell line expressing enhanced GFP fused at the N-terminus with the previously described HaloTag protein was prepared.³⁰ Additionally, a second analogous HeLa cell line was prepared containing a D106A mutation in the HaloTag part of the fusion protein to enable counterscreening for false positives. A pBabeD-based retroviral vector system was used for cell line generation (modified version of Cell Biolabs pBabe plasmid, plasmids DU29230 and DU29215, respectively; MRC-PPU Reagents and Services, https://mrcppureagents.dundee.ac.uk). In the event, HEK293-FT cells at ~60% confluence in a 10 cm dish were transfected with 6 µg of the respective pBabeD constructs, 3.8 µg of pCMV-GAG/POL (Clontech, Saint-Germain-en-Laye, France), and 2.2 µg of pCM-VSVG (Clontech) using PEI (2 µg/mL). The medium was exchanged on the next day, and after a further 24 h, virus-containing medium was harvested and cleaned using a 0.45 μm filter. HeLa cells (University of Dundee MRC-PPU stock, obtained from the American Type Culture Collection, Manassas, VA) were then transduced with the virus-containing medium supplemented with 10 μg/mL polybrene for 24 h. Cells were grown for an additional 24 h in normal growth medium and selected for 3 days in 2 µg/mL puromycin-containing medium. Finally, single-cell clones were isolated using fluorescence-activated cell sorting (FACS) based on GFP fluorescence. The cell clones selected for screening and counterscreening both exhibited similar magnitude of fluorescence intensity, although the cells containing the HaloTag D106A mutation were marginally less bright. Stably transduced HeLa cells were subsequently cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum (FBS), 2 mM GlutaMAX, 50 U/mL penicillin, 50 µg/mL streptomycin, and 1 µg/mL puromycin, at 37 °C with 5% CO₂.

Degradation Screening Assay

Samples of HaloPROTAC test compounds were transferred into black clear-base 384-well tissue culture assay plates (Greiner Bio-One, Stonehouse UK). An Echo acoustic liquid handler (Labcyte, High Wycombe, UK) was used for this procedure, with transfer volumes of 200 nL and 20 nL for 10 µM and 1 µM screening, respectively. For concentration−response curve format assays, typically a 250× test concentration dilution series of each compound was first prepared in 100% DMSO, with up to 11 concentration points differing in half-log unit increments. This series was next further diluted 25× in Fluorobrite DMEM (Thermo Fisher), and 5 µL was then transferred to the assay plate. Cultured cells were harvested at 80%−90% confluency using TrypLE enzyme; resuspended in Fluorobrite DMEM supplemented with 10% heat-inactivated FBS, 2 mM GlutaMAX, 50 U/mL penicillin, and 50 µg/mL streptomycin; and seeded into assay plates containing test compounds at a density of 10,000 cells per well. The final assay volume was 50 µL per well. DMSO concentrations in the assays were 0.4% or less. Assay plates were then incubated at 37 °C with 5% CO₂ for 2 days. Hoechst 33342 dye (0.5 µg/mL) was added 30 min before the end of the incubation, and the resulting GFP fluorescence in each well was then measured with a PHERAstar FSX microplate reader (BMG Labtech, Aylesbury, UK). Where stated (Fig. 3 and Suppl. Figs. S5, S6, and S8), GFP levels were also measured using fluorescence imaging or flow cytometry with an IN Cell Analyzer 6000 (GE Healthcare, Amersham, UK) or BD FACSCanto II (BD Biosciences, Wokingham, UK.). All assays included negative 0% effect control wells of DMSO vehicle-treated HaloTag-GFP cells and positive 100% effect control wells of untreated HeLa parental cells, and raw GFP fluorescence values from each well were converted to percent remaining or percent degradation values for all methods by normalizing to these control wells. The HaloPROTAC screening library as described above also included additional positive control wells corresponding to the known VHL ligand 6 to evaluate the in situ chemical synthesis, compound transfer, and assay performance for each 384-well plate.

Hit Selection

To select compounds for follow-up from the HaloPROTAC screening library, simple numerical filters were applied to the average (n = 3) 10 µM test concentration target degradation percentage values, and hits were defined as compounds causing greater than 50% degradation of the HaloTag-GFP fusion protein but less than 50% degradation of the HaloTag (D106A)-GFP fusion protein. To make the selection more inclusive, compounds causing greater than 40% degradation of HaloTag-GFP but less than 30% degradation of HaloTag (D106A)-GFP were also included in the hit definition. A small percentage of outlier control wells were excluded before making the percentage degradation calculations.

Results

Development of a HaloPROTAC–HaloTag-GFP Degradation 384-Well Screen and False-Positive Counterscreen

The HaloPROTAC screen and false-positive counterscreen designs are shown schematically in Figure 1B,C. To enable the development and validation of the degradation assay and the false-positive counterscreen, known positive control VHL-HaloPROTACs 7 and 8 ( Fig. 3A ) were prepared (synthetic details described in the Supplemental Material). Target protein degradation was assessed by measuring the reduction in GFP fluorescence in response to cell treatment with these VHL-HaloPROTACs in both HaloTag-GFP and HaloTag (D106A)-GFP false-positive counterscreen cell lines. The D106A HaloTag mutation abrogates the ability of the HaloTag protein to covalently bind chloroalkanes, which means that any GFP signal decrease observed in this counterscreen cell line cannot be due to on-target HaloPROTAC activity.³² As expected, treatment with VHL HaloPROTACs 7 and 8 caused a concentration-dependent reduction in fluorescence signal intensity in the HaloTag-GFP cells but had no effect on the fluorescence signal in the HaloTag (D106A)-GFP counterscreen cells ( Fig. 3 ). The fluorescence signal decrease in the HaloTag-GFP cells was more pronounced after 2 days of incubation as compared with 1 day ( Fig. 3C ), and no significant difference in the normalized response curves was evident between HeLa HaloTag-GFP clones with different HaloTag-GFP expression levels ( Suppl. Figs. S5 and S6 ).

To determine the suitability of different approaches to deploy for our screening strategy in 384-well plate format, we investigated a live cell whole-well fluorescence endpoint assay, an imaging-based endpoint assay, and a flow cytometry assay. We found that each of these methods provided approximately equivalent response curves, good reproducibility between technical replicates, and high Z′ values ( Suppl. Fig. S8 ). These studies indicated that there were multiple robust, fit-for-purpose options to use for the detection of true positive degraders as well as false-positive results that might occur due to growth inhibition, cytotoxicity, or other nonspecific effects. For the subsequent screening, brightly fluorescent HeLa HaloTag-GFP and HaloTag (D106A)-GFP mutant cell clones with normal HeLa morphology and growth rate were selected for use, along with a 2-day assay incubation time. For data acquisition, we selected the whole-well fluorescence endpoint method (PHERAstar microplate reader) because of its efficiency and good performance, and the positive control VHL-HaloPROTAC building block amine 6 ( Fig. 3A ) was employed as a per-plate standard to validate and control the consistency of the in situ chemistry, the acoustic compound transfer efficiency, and the overall assay performance for each 384-well plate.

High-Throughput Chemistry Optimization and Generation of a HaloPROTAC Compound Screening Library

As described above, we designed a simple, one-stage coupling protocol between a collection of diverse amine building blocks 2 and a single primary alkyl chloride NHS-activated ester building block 4 ( Fig. 2 ) to enable the preparation of a HaloPROTAC library in plate-based format. To optimize the reaction conditions for efficient coupling, a range of bases, reaction times, and reagent stoichiometry were evaluated employing the VHL-HaloPROTAC amine building block 6 ( Fig. 3A ) as a prototypical amine for array chemistry.

Extensive experimentation demonstrated that robust and reproducible coupling was obtained using DMSO as a solvent with a 1:1 molar ratio of amine 2 to NHS ester 4, 10 equivalents of NMM as base, and a reaction time of 16 h. Importantly, all reaction components were evaluated in our HaloTAG-GFP fluorescence assay system and were not observed to cause any detrimental effects on the cells or any decrease in fluorescence signal ( Suppl. Fig. S7 ). We also observed that the HaloTag-GFP fluorescence assay results generated with in situ prepared HaloPROTAC control compound 7 gave a similar result to those generated using HaloPROTAC control compound 7 that was isolated and purified prior to testing ( Suppl. Fig. S7 ), suggesting that isolation and purification of the HaloPROTAC library compounds were not required prior to testing in cells. During the course of reaction optimization, it was discovered that airflow-mediated evaporation of NMM in a standard fume hood was associated with substantially incomplete reactions, especially in the corners and along the edges of the plates. To mitigate the poor reaction conversion due to evaporative loss of NMM, we elected to conduct all subsequent plate-based chemistry in a glovebox under a positive-pressure nitrogen atmosphere.

Following plate-based chemistry optimization, we next prepared the screening library of HaloPROTAC test molecules ( Fig. 2B ). We selected a total of 2934 amines for use as building blocks for HaloPROTAC library synthesis, and employing our optimized reaction conditions, we observed >50% conversion to product for 64% of amines, 20%−50% conversion for 6% of amines, and <20% conversion for 30% of amines ( Fig. 2C ). Factors influencing reaction conversion appear to be both physicochemical ( Suppl. Fig. S3 ) and steric ( Fig. 2D ) in nature, with steric effects dominating the reaction outcome. Successful conversion rates decreased in the order of secondary cyclic nonhindered > primary nonhindered > primary hindered > secondary linear nonhindered > secondary cyclic hindered > secondary linear hindered amines (Fig. 2D and Suppl. Fig. S2). With more than 1900 library compounds successfully prepared with over 50% conversion via high-throughput chemistry, the plates were then assessed in the phenotypic screening assays.

Screening of HaloPROTAC Compound Library for HaloTag-GFP Degrading Compounds and False-Positive Counterscreening

The crude HaloPROTAC compounds at 10 µM concentration were tested for their ability to induce a reduction in GFP fluorescence in the HaloTag-GFP cells and in the HaloTag (D106A)-GFP counterscreen cells (n = 3, tested on different days). Screening was also performed at 1 µM concentration (n = 1) using the HaloTag-GFP cells in an attempt to detect more potent degraders. The hit activity histograms are shown in Figure 4 . The positive control VHL-HaloPROTAC 7, which was prepared in situ in 16 wells on each assay plate, consistently caused a high level of fluorescence reduction (73% average across all control wells) in the HaloTag-GFP cells but had minimal effect on fluorescence in the HaloTag (D106A)-GFP counterscreen cells. We found that a significant proportion of the tested HaloPROTAC screening set compounds caused some reduction in HaloTag-GFP fluorescence, with 6% of test compound wells reducing GFP levels by more than 50%. However, the activity histogram for the HaloTag-GFP cells was similar to that for the HaloTag (D106A)-GFP counterscreen cells, indicating that many of the HaloPROTAC compounds inducing a reduction in GFP fluorescence were likely to be false positives that were not acting through a targeted degradation mechanism.

Figure 4.

Results of screening the HaloPROTAC set. (A−C) Hit activity distribution histograms. (D) Comparison of screening and counterscreening results using HaloTag-GFP and HaloTag (D106A)-GFP cells, respectively. Line is y = x. One outlier with a high raw fluorescence value is excluded from the figure. The numerical percent activity filters used to select hits are shown.

To identify the potential functional degraders from the many apparent false-positive hits, we compared the average effects of 10 µM HaloPROTAC compounds in the screening and counterscreening cells (mean, n = 3) ( Fig. 4D ) and applied simple thresholds based on the percentage decrease in the GFP fluorescence signal, as described in Materials and Methods. This resulted in nine HaloPROTAC compounds being identified as potential true degrader hits. These nine compounds were then individually re-prepared from stock solutions in the same manner as described above and retested in a concentration−response format using both the screening and counterscreening cell lines. We observed that two of these nine compounds (HaloPROTAC A and HaloPROTAC B) gave reproducibly (n = 2) differentiated concentration−response profiles between the HaloTag-GFP and the control HaloTag (D106A)-GFP cell lines that were consistent with the response expected for a specific HaloTag-GFP degrader ( Fig. 5 ). To further evaluate these two putative hit compounds, we resynthesized HaloPROTAC A and HaloPROTAC B again, this time starting from the de novo syntheses of the amine building blocks, and we also included a separate cell viability assay in the testing protocol. Following testing of these newly resynthesized HaloPROTAC hits in both the screening and counterscreening cell lines, we observed that HaloPROTAC A appeared to be a cytotoxic false-positive hit, causing a significant reduction in the intracellular ATP levels in both test cell lines. Surprisingly, we observed that this compound was more cytotoxic against the HaloTag-GFP HeLa cells than against the HaloTag (D106A)-GFP HeLa cells ( Suppl. Fig. S9 ), which was a confounding result leading to the initial classification of this compound as a true degrader. Unfortunately, we were unable to reproduce the previously observed HaloTag-GFP degradation effect with the re-prepared sample of HaloPROTAC B, and there was not enough of the original sample remaining to undertake further structural or biological analysis.

Figure 5.

Hit confirmation in concentration−response format showing reproducible differential HaloTag-GFP versus counterscreen HaloTag (D106A)-GFP fluorescence reduction profiles for the two out of nine hits that were confirmed.

Discussion

Targeted protein degradation is a field with significant potential for industrial and academic research applications, and the first examples of PROTACs entering clinical trials have recently been described.^{5,8,12,33−35} Currently reported PROTAC molecules employ a limited catalog of E3 ligases to induce ubiquitin transfer and subsequent target degradation, and increasing the available E3 ligase repertoire for PROTAC design represents a clear opportunity to expand the scope of PROTAC utility for medicine discovery.²⁶ However, this goal is challenging to realize due to difficulties typically associated with the expression, purification, and screening of biologically relevant E3 ligase complexes. Consequently, limited examples of successful high- or medium-throughput screening approaches for E3 ligases have been reported to date.^36–39

In contrast to a traditional affinity binding or biochemical E3 ligase screening campaign, we envisaged that a phenotypic approach might provide a straightforward solution for the identification of molecules capable of effecting target degradation in cells.⁴⁰ Such an approach would have the additional advantage of multiplexed target screening, since the entire cellular repertoire of E3 ligases and other protein degradation machinery would be sampled simultaneously, and we reasoned that a plate-based HaloPROTAC array could provide a good opportunity to implement this approach.⁴¹ HaloPROTACs employ an E3 ligase binder with a linker bearing a primary chloroalkane, and they have been previously demonstrated to covalently bind to HaloTag-POI fusion proteins, resulting in ubiquitin transfer and subsequent degradation of the HaloTag-POI fusion. To the best of our knowledge, the HaloPROTAC concept has not been previously used for screening purposes.

To enable our screening strategy, we generated a clonal HeLa cell line expressing the HaloTag-GFP fusion protein, and we tested the performance of these cells with the known VHL HaloPROTACs 7 and 8. Since we anticipated that many off-target effects unrelated to proteasomal target degradation could result in decreases in the HaloTag-GFP fluorescence signal, we also generated a clonal cell line containing a mutant HaloTag (D106A)-GFP that is unable to covalently bind to HaloPROTAC molecules. This false-positive counterscreening cell line was subsequently used to identify compounds causing loss of fluorescence signal due to mechanisms unrelated to protein degradation. We also incorporated an ATP-based cellular viability assay into the follow-up screening protocol to eliminate potentially cytotoxic compounds.

Chimeric molecules such as PROTACs are not widely represented in typical screening libraries, and the ability to access a library of HaloPROTAC degraders is envisaged to be useful for drug discovery and chemical biology purposes. To this end, we developed a high-throughput chemistry protocol in 384-well plate format to combine a diverse selection of amine building blocks with a single linker bearing a primary chloroalkane HaloTag recognition element. During library construction, we observed conversion rates to product HaloPROTACs of >50% for approximately 64% of the starting amine building block library, which was deemed to be fit for purpose for the evaluation of an initial degradation screen. We also identified steric and physicochemical parameters such as basicity, solubility, size, and flexibility of the amines to be important for controlling the reaction rate and ultimate reaction success, which are envisioned to be useful for future library design and preparation.

Screening of the HaloPROTAC library in the HaloTag-GFP HeLa cells as well as counterscreening in the HaloTag (D106A)-GFP cells revealed that many compounds significantly reduced the fluorescence signal in both cell lines. Because the HaloPROTACs are unable to bind the HaloTag D106A mutant,⁴² we speculate that compounds reducing fluorescence in both cell lines are acting primarily through cytotoxicity or other nuisance mechanisms affecting the proliferation and viability of the cells during the 48 h assay incubation, rather than through a true on-target degradation mechanism. This effect demonstrates the well-established need to accurately discern false positives in phenotypic screening campaigns.

We identified nine HaloPROTAC compounds as potential true HaloTag-GFP degrader hits worthy of progression to confirmatory concentration–response testing. Out of these nine compounds, only HaloPROTACs A and B were found to cause reproducible concentration-dependent reduction in HaloTag-GFP fluorescence with a clear window when compared with reduction in the control HaloTag (D106A)-GFP cell line fluorescence. However, upon de novo resynthesis, neither of these compounds was found to be a genuine HaloTag-GFP degrader that qualified for further investigation.

The identification of HaloPROTAC A as an initial hit appears to have been due to the differential cytotoxicity/cell viability potency of this compound between the HeLa HaloTag GFP screening cells and the counterscreen HeLa HaloTag (D106A)-GFP cell clones ( Suppl. Fig. S9 ). Although surprising, this result is not inconsistent with previously reported variability that has been observed between different HeLa clonal cell lines.⁴³ In contrast, our inability to reproduce HaloPROTAC B-mediated GFP degradation following de novo resynthesis of HaloPROTAC B highlights an important caveat of following up hits directly from compound library stock solutions and emphasizes the need to re-prepare hits on an individual de novo basis for assay confirmation. Unfortunately, depletion of the original compound stock solution prevents us from conducting a rigorous structural analysis of the library building blocks used for the preparation of HaloPROTAC B. Therefore, the actual structural identity of the HaloPROTAC B screening hit is unknown.

Since effective target degradation is dependent on many factors, we speculate that this screening approach can be further improved not only through the addition of more amine components but also by employing additional linker building blocks with different lengths and compositions to increase the odds of effective ternary complex formation leading to degradation. Furthermore, screening HaloPROTAC libraries in different cell lines may also increase the odds of detecting functional degraders, since different cell lines can have differential expression of degradation machinery component proteins. Future iterations of this screening design might also benefit from other established counterscreening approaches, such as co-expression of the HaloTag-GFP with a second fluorescent or bioluminescent control protein.^6,44–47

Despite our inability to identify novel functional HaloPROTACs with on-target HaloTag-GFP degradation in this study, we believe that our phenotypic screening approach is fit for purpose for the identification of new chemical matter capable of recruiting E3 ligases or other elements of the cellular protein degradation machinery in an unbiased way and will be a valuable new method for the discovery of target protein degraders.

Supplemental Material

sj-pdf-1-jbx-10.1177_24725552211017517 – Supplemental material for A Phenotypic Approach for the Identification of New Molecules for Targeted Protein Degradation Applications

Supplemental material, sj-pdf-1-jbx-10.1177_24725552211017517 for A Phenotypic Approach for the Identification of New Molecules for Targeted Protein Degradation Applications by Peter Stacey, Hannah Lithgow, Xiao Lewell, Agnieszka Konopacka, Stephen Besley, Georgina Green, Ryan Whatling, Robert Law, Sascha Röth, Gopal Sapkota, Ian E. D. Smith, Glenn A. Burley, John Harling, Andrew B. Benowitz, Markus A. Queisser and Marcel Muelbaier in SLAS Discovery

Footnotes

Acknowledgements

The authors are grateful to Richard Kasprowicz for help with the analysis of fluorescence images and to Blandine McKay and David Battersby for helpful discussions.

Supplemental material is available online with this article.

Declaration of Conflicting Interests

The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: PS, HL, XL, AK, SB, GG, RL, IEDS, JH, ABB, MAQ, and MM are current or former employees and/or shareholders of GlaxoSmithKline, and their research and authorship of this article were completed within the scope of their employment. RW is a former industrial placement student at GlaxoSmithKline. The other authors declare no potential conflicts of interest with respect to the research, authorship, or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was funded by GlaxoSmithKline PLC, GSK/University of Strathclyde Centre for Doctoral Training in Synthetic and Medicinal Chemistry, EPSRC via Prosperity Partnership EP/S035990/1, and UK MRC grant number MC_UU_00018/6.

ORCID iDs

Robert Law

Glenn A. Burley

References

Burslem

G. M.

Crews

C. M.

Proteolysis-Targeting Chimeras as Therapeutics and Tools for Biological Discovery. Cell 2020, 181, 102–114.

Lai

A. C.

Crews

C. M.

Induced Protein Degradation: An Emerging Drug Discovery Paradigm. Nat. Rev. Drug Discov. 2017, 16, 101–114.

Churcher

Protac-Induced Protein Degradation in Drug Discovery: Breaking the Rules or Just Making New Ones?

J. Med. Chem. 2018, 61, 444–452.

Bondeson

D. P.

Mares

Smith

I. E.

, et al. Catalytic In Vivo Protein Knockdown by Small-Molecule PROTACs. Nat. Chem. Biol. 2015, 11, 611–617.

Nowak

R. P.

Jones

L. H.

Target Validation Using PROTACs: Applying the Four Pillars Framework. SLAS Discov. 2021, 26, 474–483.

Kastl

J. M.

Davies

Godsman

, et al. Small-Molecule Degraders beyond PROTACs—Challenges and Opportunities. SLAS Discov. 2021, 26, 524–533.

Zhang

Loh

Chen

, et al. Targeted Protein Degradation Mechanisms. Drug Discov. Today Technol. 2019, 31, 53–60.

Luh

L. M.

Scheib

Juenemann

, et al. Prey for the Proteasome: Targeted Protein Degradation—A Medicinal Chemist’s Perspective. Angew Chem. Int. Ed. Engl. 2020, 59, 15448–15466.

Gadd

M. S.

Testa

Lucas

, et al. Structural Basis of PROTAC Cooperative Recognition for Selective Protein Degradation. Nat. Chem. Biol. 2017, 13, 514–521.

10.

Collins

G. A.

Goldberg

A. L.

The Logic of the 26S Proteasome. Cell 2017, 169, 792–806.

11.

Verma

Mohl

Deshaies

R. J.

Harnessing the Power of Proteolysis for Targeted Protein Inactivation. Mol. Cell 2020, 77, 446–460.

12.

Mullard

Arvinas’s PROTACs Pass First Safety and PK Analysis. Nat. Rev. Drug Discov. 2019, 18, 895.

13.

Mullard

Targeted Protein Degraders Crowd into the Clinic. Nat. Rev. Drug Discov. 2021, 20, 247–250.

14.

Bulatov

Ciulli

Targeting Cullin-RING E3 Ubiquitin Ligases for Drug Discovery: Structure, Assembly and Small-Molecule Modulation. Biochem. J. 2015, 467, 365–386.

15.

Zengerle

Chan

K. H.

Ciulli

Selective Small Molecule Induced Degradation of the BET Bromodomain Protein BRD4. ACS Chem. Biol. 2015, 10, 1770–1777.

16.

Qian

Altieri

, et al. Hijacking the E3 Ubiquitin Ligase Cereblon to Efficiently Target BRD4. Chem. Biol. 2015, 22, 755–763.

17.

Tomoshige

Hashimoto

Ishikawa

Efficient Protein Knockdown of HaloTag-Fused Proteins Using Hybrid Molecules Consisting of IAP Antagonist and HaloTag Ligand. Bioorg. Med. Chem. 2016, 24, 3144–3148.

18.

Mares

Miah

A. H.

Smith

I. E. D.

, et al. Extended Pharmacodynamic Responses Observed upon PROTAC-Mediated Degradation of RIPK2. Commun. Biol. 2020, 3, 140.

19.

Hines

Lartigue

Dong

, et al. MDM2-Recruiting PROTAC Offers Superior, Synergistic Antiproliferative Activity via Simultaneous Degradation of BRD4 and Stabilization of p53. Cancer Res. 2019, 79, 251–262.

20.

Chamberlain

P. P.

Hamann

L. G.

Development of Targeted Protein Degradation Therapeutics. Nat. Chem. Biol. 2019, 15, 937–944.

21.

Martin-Acosta

Xiao

PROTACs to Address the Challenges Facing Small Molecule Inhibitors. Eur. J. Med. Chem. 2021, 210, 112993.

22.

Schapira

Calabrese

M. F.

Bullock

A. N.

, et al. Targeted Protein Degradation: Expanding the Toolbox. Nat. Rev. Drug Discov. 2019, 18, 949–963.

23.

Madsen

Azevedo

Micco

, et al. An Overview of DNA-Encoded Libraries: A Versatile Tool for Drug Discovery. Prog. Med. Chem. 2020, 59, 181–249.

24.

Goodnow

R. A.

Jr. Dumelin

C. E.

Keefe

A. D.

DNA-Encoded Chemistry: Enabling the Deeper Sampling of Chemical Space. Nat. Rev. Drug Discov. 2017, 16, 131–147.

25.

Zhu

Grady

L. C.

Ding

, et al. Development of a Selection Method for Discovering Irreversible (Covalent) Binders from a DNA-Encoded Library. SLAS Discov. 2019, 24, 169–174.

26.

Ishida

Ciulli

E3 Ligase Ligands for PROTACs: How They Were Found and How to Discover New Ones. SLAS Discov. 2021, 26, 484–502.

27.

Goodnow

Jr.

DNA-Encoded Library Technology (DELT) after a Quarter Century. SLAS Discov. 2018, 23, 385–386.

28.

Simard

J. R.

Lee

Vieux

, et al. High-Throughput Quantitative Assay Technologies for Accelerating the Discovery and Optimization of Targeted Protein Degradation Therapeutics. SLAS Discov. 2021, 26, 503–517.

29.

Los

G. V.

Encell

L. P.

McDougall

M. G.

, et al. HaloTag: A Novel Protein Labeling Technology for Cell Imaging and Protein Analysis. ACS Chem. Biol. 2008, 3, 373–382.

30.

Buckley

D. L.

Raina

Darricarrere

, et al. HaloPROTACS: Use of Small Molecule PROTACs to Induce Degradation of HaloTag Fusion Proteins. ACS Chem. Biol. 2015, 10, 1831–1837.

31.

Boss

Hazemann

Kimmerlin

, et al. The Screening Compound Collection: A Key Asset for Drug Discovery. Chimia (Aarau) 2017, 71, 667–677.

32.

Neklesa

T. K.

Crews

C. M.

Chemical Biology: Greasy Tags for Protein Removal. Nature 2012, 487, 308–309.

33.

Roth

Macartney

T. J.

Konopacka

, et al. Targeting Endogenous K-RAS for Degradation through the Affinity-Directed Protein Missile System. Cell Chem. Biol. 2020, 27, 1151–1163.e6.

34.

Gao

Sun

Rao

PROTAC Technology: Opportunities and Challenges. ACS Med. Chem. Lett. 2020, 11, 237–240.

35.

Ding

Fei

Emerging New Concepts of Degrader Technologies. Trends Pharmacol. Sci. 2020, 41, 464–474.

36.

Landre

Rotblat

Melino

, et al. Screening for E3-Ubiquitin Ligase Inhibitors: Challenges and Opportunities. Oncotarget 2014, 5, 7988–8013.

37.

Foote

P. K.

Statsyuk

A. V.

Monitoring PARKIN RBR Ubiquitin Ligase Activation States with UbFluor. Curr. Protoc. Chem. Biol. 2018, 10, e45.

38.

Rossi

Rotblat

Ansell

, et al. High Throughput Screening for Inhibitors of the HECT Ubiquitin E3 Ligase ITCH Identifies Antidepressant Drugs as Regulators of Autophagy. Cell Death Dis. 2014, 5, e1203.

39.

Tian

Zeng

Liu

, et al. A Cell-Based High-Throughput Screening Method Based on a Ubiquitin-Reference Technique for Identifying Modulators of E3 Ligases. J. Biol. Chem. 2019, 294, 2880–2891.

40.

Mayor-Ruiz

Bauer

Brand

, et al. Rational Discovery of Molecular Glue Degraders via Scalable Chemical Profiling. Nat. Chem. Biol. 2020, 16, 1199–1207.

41.

Heidary

D. K.

Fox

Richards

C. I.

, et al. A High-Throughput Screening Assay Using a Photoconvertable Protein for Identifying Inhibitors of Transcription, Translation, or Proteasomal Degradation. SLAS Discov. 2017, 22, 399–407.

42.

Neklesa

T. K.

Tae

H. S.

Schneekloth

A. R.

, et al. Small-Molecule Hydrophobic Tagging-Induced Degradation of HaloTag Fusion Proteins. Nat. Chem. Biol. 2011, 7, 538–543.

43.

W. E.

Zhang

Guo

Q. F.

, et al. HeLa-CCL2 Cell Heterogeneity Studied by Single-Cell DNA and RNA Sequencing. PLoS One 2019, 14, e0225466.

44.

Wolff

N. C.

Pavía-Jiménez

Tcheuyap

V. T.

, et al. High-Throughput Simultaneous Screen and Counterscreen Identifies Homoharringtonine as Synthetic Lethal with Von Hippel-Lindau Loss in Renal Cell Carcinoma. Oncotarget 2015, 6, 16951–16962.

45.

Zanella

Lorens

J. B.

Link

High Content Screening: Seeing Is Believing. Trends Biotechnol. 2010, 28, 237–245.

46.

Koren

Timms

R. T.

Kula

, et al. The Eukaryotic Proteome Is Shaped by E3 Ubiquitin Ligases Targeting C-Terminal Degrons. Cell 2018, 173, 1622–1635.e14.

47.

Paguio

Stecha

Wood

K. V.

, et al. Improved Dual-Luciferase Reporter Assays for Nuclear Receptors. Curr. Chem. Genomics 2010, 4, 43–49.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

1.29 MB