Demonstrating In-Cell Target Engagement Using a Pirin Protein Degradation Probe (CCT367766)

Demonstrating intracellular protein target engagement is an essential step in the development and progression of new chemical probes and potential small molecule therapeutics. However, this can be particularly challenging for poorly studied and noncatalytic proteins, as robust proximal biomarkers are rarely known. To confirm that our recently discovered chemical probe 1 (CCT251236) binds the putative transcription factor regulator pirin in living cells, we developed a heterobifunctional protein degradation probe. Focusing on linker design and physicochemical properties, we generated a highly active probe 16 (CCT367766) in only three iterations, validating our efficient strategy for degradation probe design against nonvalidated protein targets.


■ INTRODUCTION
Drug discovery is reliant on recombinant proteins and biochemical screens to develop structure−activity relationships (SAR) and progress compounds. 1 However, the conditions in biochemical assays often display little relevance to the intracellular environment, which can result in a failure to translate high target affinity to activity within a living cell. 2 Target-proximal biomarker modulation is the most important confirmation of intracellular target engagement. 3 Unfortunately, this is often not possible in early stage chemical probe or drug discovery projects, especially against novel biological targets, where poorly understood biology and pharmacology make it difficult to discover and robustly validate biomarkers, a process that is particularly challenging for noncatalytic proteins.
Owing to the importance of confirming intracellular target engagement, several techniques have been developed. The overexpression of fusion proteins allows for a direct readout of target occupancy. 4 However, the engineered cell lines are often difficult to generate and the protein label can impact compound binding. The cellular thermal shift assay (CETSA) is a label-free technique for intracellular target engagement. 5 It exploits compoundinduced stabilization of protein melting temperatures, but CETSA cannot be applied to all targets. 6 Activity-based protein profiling (ABPP) methods utilize nonselective irreversible covalent ligands, or intracellular fluorescence polarization probes, combined with intracellular reversible ligand competition, to study target engagement. 7 Proteolysis targeting chimeras (PROTACs) 8,15 and specific and nongenetic IAP-dependent protein erasers (SNIPERs) 9 are heterobifunctional molecules that induce rapid and selective protein degradation, via the proteasome, within living cells. One portion of the molecule engages the target protein, while the other, attached via a flexible linker, recruits an E3 ligase to ubiquitinate the target, marking it for degradation as part of the cullin−RING finger machinery. 10,15 We recently reported the development of a high affinity pirin chemical probe 1 (CCT251236, SPR K D = 44 nM) discovered through a cell-based phenotypic screen for inhibitors of the heat shock transcription factor 1 (HSF1) stress pathway. 11 Pirin is an iron-binding member of the cupin super family of proteins and has been reported as a putative transcription factor regulator. 12,13 It has no known enzymatic function in mammalian cells, no endogenous ligands have been reported, and no validated proximal biomarkers have been described. 14 This makes demonstrating intracellular target engagement in living cells very challenging.
We hypothesized that we could demonstrate chemical probe 1 binding to pirin within living cells by developing a pirin-targeting protein degradation probe (PDP).

■ RESULTS AND DISCUSSION
Protein Degradation Probe Design. PDPs have been described against various proteins, 15 with the bromodomain epigenetic target, BRD4, the most extensively studied. 16 These heterobifunctional PDP molecules have utilized several ligands that bind to the E3 ligases: VHL, 17 the IAP proteins, 18 and CRBN. 19 Despite the rapid expansion in PDP research, there remains no clear methodology to determine which proteins are amenable to PDP-mediated degradation, 20 which E3 ligase ligand should be exploited or the optimal features of probe design. The structure of the linker, its length, and physicochemical properties have all been demonstrated to be important for PDP activity. 21 The linker controls the formation of the essential ternary complex, 22 with evidence that it may stabilize the protein−protein interaction (PPI) between the target and the E3 ligase rather than forming a detached linear ternary complex ( Figure 1), although it is unclear whether the ternary complex is part of the multiprotein cullin−RING finger complex. 23−25 Although we had discovered a high affinity ligand for pirin, there was no evidence that this protein would be a suitable substrate for ubiquitination by a PDP-recruited E3 ligase, without which extensive linker optimization could be futile. Therefore, we initially designed a synthetically tractable 15-atom linker that we predicted would not affect the affinity of the PDP for the isolated target proteins. 26 Analysis of the crystal structure of the chemical probe 1 bound to pirin (Figure 1, PDB 5JCT) 11 suggested that the solvent-exposed solubilizing group vector should be amenable to linker attachment. We selected a CRBN-targeting thalidomide ligand as the basis of our E3 ligase binding motif due to its low molecular weight. The CRBN-targeting ligand would be attached via the solvent exposed hydroxyl group of the 4-hydroxythalidomide analogue 2. 27 The linker would then attach the pirin-and CRBN-targeting specific binding groups via two amide moieties to give our first generation pirin-targeting PDP 3 (Scheme 1).
The CRBN-targeting motif 4 of PDP 3 was synthesized from 4-hydroxythalidomide 2 in two steps and 84% yield, in a similar manner to that previously described. 28 The pirin-binding motif of the PDP 3 was synthesized from 6-bromoquinoline 5 via a palladium-mediated carbonylation reaction on the ether derivative 6. Trapping the carbonylation intermediate with 2-(trimethylsilyl)ethan-1-ol gave the (trimethylsilyl)ethyl ester 7, 29 which facilitated TBAF-selective hydrolysis. Amide coupling to the previously described bis-aniline derivative 8 11 was followed by hydrolysis of the aliphatic linker ester to give acid 9. Final amide coupling to the CRBN-targeting derivative 4 gave the first-generation pirin-targeting PDP 3 in 10 steps and 0.4% overall yield.
The First Generation PDP. Analysis of the heterobifunctional PDP 3 revealed that it possessed good affinity for recombinant Figure 1. (top) Pirin (5JCT)/CRBN (5CI1)/PDP ternary complex design model. The PDP can either stabilize a PPI or simply bring the proteins in close proximity depending on the role of the linker. (bottom left) Cartoon representation of the chemical probe 1 (yellow) bound to recombinant pirin (5JCT). The cloud represents the shape of the binding pocket with key residues shown in black and the metal in orange. Red = oxygen, blue = nitrogen. Hydrogens and solvent are omitted for clarity except the water coordinated to the metal, shown as a red sphere. Both representations were generated using the PyMOL Molecular Graphics System, version 1.8; Schrodinger, LLC. (bottom right) Key residues in the binding site and the clear solvent exposed vector for the chemical probe 1 binding to pirin are shown adapted from an analysis using MOE 2014.09. The ethyl pyrrolidine solubilizing group of chemical probe 1 was not resolved in the crystal structure and therefore is not drawn in the analysis. pirin ( Table 1, entry 1) when measured using SPR, confirming the success of the rationally designed attachment vector. 30 The affinity of PDP 3 was then assessed against the CRBN-DDB1 complex, with DDB1 acting as a scaffolding protein, using an FP-assay similar to that previously described (Supporting Information, Figure S6). 31 PDP 3 displayed moderate affinity for CRBN-DDB1, with K i = 230 nM, 32 comparable to the affinities of the parent CRBN ligands, thalidomide and lenalidomide (Supporting Information, Figure S7).
Following confirmation that both binding motifs of the PDP 3 retained high affinity for their respective targets, we then investigated its activity against pirin in human cancer cell lines.
Several cell lines were assessed for CRBN expression by quantitative capillary electrophoresis (Supporting Information, Figure S17). 33 The SK-OV-3 ovarian carcinoma cell line 34 displayed good basal CRBN and also pirin expression, and there was no observable depletion of pirin in these cells when treated with chemical probe 1 (1 μM, data not shown), so this line was selected for further study. Unfortunately, treatment of SK-OV-3 cells with PDP 3, at high concentrations (>1 μM) and for extended time periods (>48 h), resulted in no measurable effects on the cancer cells (data not shown).
The Second Generation PDP. We speculated several causes for the failure of the first generation PDP 3. Pirin may simply be Table 1. Physicochemical Properties and Affinities for Recombinant Protein Targets of the Three Generations of PDPs I a HBD = hydrogen bond donor count. b ALogP was calculated using Biovia Pipeline Pilot, version 9.5, 2 SF. c Log D 7.4 measured using a HPLC-based method, n = 1, 2 SF. d tPSA was calculated using ChemDraw (16.0.1.4) based on the O-and N-count, 3 SF. e KS = kinetic solubility in pH 7.4 phosphate buffer at room temperature, n = 1, 1 SF. f K D values are reported to 2 SF and are calculated by equilibrium analysis using a one site specific binding model from SPR sensorgrams at equilibrium where possible, pK D = −log(K D (M) × 10 −9 ) and represents the geometric mean of n = 3 independent biological repeats. g IC 50 values are reported to 2 SF and are calculated from an FP-assay dose−response curve to displace a thalidomide derived fluorescent probe using a log[inhibitor] vs response − variable slope (four parameters) model, pIC 50 = −log(IC 50 (M) × 10 −9 ) and represents the geometric mean of n = 3 independent biological repeats, also see ref 31. h  incompatible with this methodology, and if so, no further improvements could be made. 35 Alternatively, CRBN could be the wrong E3 ligase target to deplete pirin, 36 or the linker length could be inconsistent with formation of the ternary complex. 26 Finally, the physicochemical properties of PDP 3 could be limiting its intracellular free concentration. 37 Given these variables, designing a PDP against a protein target that has not previously been validated as being susceptible to E3 ligase-directed degradation is challenging, as optimization cycles can appear lengthy and the difficult and low yielding synthesis discourages the generation of multiple analogues. It was also unclear how strict and narrow the requirements for optimal PDP design would be. We hypothesized that the physicochemical properties of the PDP were the primary cause of the failure of the first generation probe, and so began a redesign process that should increase cell membrane flux while maintaining the same linker length and CRBN-targeting ligand (Table 1). By carrying out multiple changes to the PDP, we aimed to minimize the number of iterative design cycles, so we could more rapidly validate this approach.
The design of heterobifunctional molecules unavoidably results in high molecular weight compounds, making it difficult to balance their physicochemical properties in a manner consistent with acceptable permeability and solubility. 38 In the case of the first generation pirin-targeting PDP 3, although the Log D 7.4 was acceptable (Table 1, entry 1), 39 the linker had introduced two new hydrogen bond donors (HBD), which can have a negative impact on permeability. The calculated tPSA was very high (258 Å 2 ), although it is inevitable that tPSA will be high for large molecules and outside the standard cut-offs for cell

Journal of Medicinal Chemistry
Article permeability unless the compounds are highly lipophilic. 40 In designing a second generation PDP, we followed standard medicinal chemistry principles to reduce the tPSA and HBD count while maintaining an acceptable Log D 7.4 . To achieve this, we redesigned the ether linker in the first generation PDP 3 to include a methylene piperazine that would project into solvent based on analysis of the crystal structure of the chemical probe 1 bound to pirin. The resulting tertiary amide would remove one HBD. We also sought to mask the quinoline amide HBD, using a dipole−dipole interaction, via a bioisosteric replacement with fluorine. 41 These changes resulted in the design of the second generation pirin-targeting PDP 10 (Scheme 2) with a modestly reduced tPSA (244 Å 2 ).
The synthesis of PDP 10 began from the fluoroaniline carboxamide 11, synthesized in a similar manner to that previously described from 2-fluoro-5-nitroaniline 12 in three steps and 85% yield. Amide coupling to give bisamide 13 and benzylic oxidation using selenium dioxide 42 was followed by reductive amination of the resulting aldehyde with N-Boc-piperazine, and subsequent deprotection, to give 14. The CRBN-targeting thalidomide derivative precursor 15 was prepared in four steps and 27% yield from 4-hydroxythalidomide 2. Final amide coupling with monosubstituted piperazine 14 gave the second generation pirintargeting PDP 10 in 11 steps and 8% overall yield (Scheme 2). PDP 10 displayed a similar affinity for recombinant pirin and CRBN-DDB1 to our first generation probe 3 (Table 1, entry 2), so it was progressed to cellular assessment of pirin degradation. Pleasingly, treatment of SK-OV-3 ovarian cancer cells with 10 at 3.0 μM total concentration for up to 48 h revealed a clear and time-dependent depletion of intracellular pirin expression (Supporting Information, Figure S18). 43 This confirmed, for the first time, that pirin is amenable to modulation using a PDP and that the bisamide chemotype not only binds recombinant pirin with high affinity but also binds pirin within living cells.
Although demonstrating pirin depletion with the second generation probe 10 was a very encouraging result, we found the effects were poorly reproducible. The concentrations of PDP 10 needed to observe pirin degradation were high and close to its kinetic solubility (KS). Furthermore, at least 24 h of compound exposure was needed before pirin degradation was observed (Supporting Information, Figure S18). To explore the cause of the variable results obtained with PDP 10, we assessed its chemical stability. At room temperature, PDP 10 was stable as a solid and in DMSO stock solution (>1 month, data not shown), but at 37°C in pH 7.4 phosphate buffer, consistent with the cell assay conditions, it underwent rapid decomposition (Supporting Information, Table S1), displaying a half-life of only ∼4 h. This poor chemical stability was consistent with the known decomposition of the parent CRBN-targeting thalidomide ligand under these conditions, where multiple hydrolysis products of the imide and glutaramide moieties are observed. 44 The chemical probe 1 displayed no instability under these conditions; therefore, we concluded that the facile hydrolysis of the CRBN-targeting motif limited the reproducibility of our slow-acting second generation pirin-targeting PDP 10.
The Third Generation PDP. In the design of a third generation probe, we aimed to increase permeability further. This should result in higher intracellular free concentrations more quickly, mitigating the poor stability of the CRBN-targeting motif. We decided to carry out a bioisosteric replacement of the central ring fluorine for the larger and more sterically hindering chlorine substituent. 45 However, we were concerned that the increase in lipophilicity from this exchange would negatively impact both the solubility and permeability of the probe. Large lipophilic heterobifunctional molecules are particularly susceptible to aggregation, 46 and this decreases the free concentration that drives cell membrane flux. 47 To balance the lipophilicity, we removed the tertiary amide bond to the piperazine, introducing a cationic amine, which would be substantially charged at pH 7.4 (MoKa, version 2.5.2, pK a = 8.0). 48 The second amide in the linker was also removed, reducing the HBD count further and increasing the overall flexibility of the ligand, consistent with the formation of the crucial PDP ternary protein complex. The linker length was reduced by one atom to accommodate these changes and resulted in the design of the third generation pirin-targeting PDP 16 (CCT367766), which now displayed a notably reduced, but still high, tPSA (207 Å 2 ).
The synthesis of the pirin-targeting motif of the third generation PDP 16 was carried out in a similar manner to that previously described starting from 2-chloro-5-nitroaniline 17 to give chlorobisamide 18 in three steps and 54% yield (Scheme 2). Oxidation of the methylquinoline moiety with SeO 2 and reductive amination of the resulting aldehyde with N-Boc-piperazine and N-Boc deprotection gave 19 in 32% yield. Following S N 2 alkylation with the ether linker to give 20, selective Mitsunobu alkylation with 4-hydroxythalidomide 2 gave the third generation PDP 16 in eight steps and 2% overall yield. A nonpirin-binding negative control matched pair compound, PDP 21, based on our negative control pirin chemical probe, 11 was synthesized from regioisomer 23 in a similar manner in 2% overall yield. A non-CRBN binding control 22 49 was also synthesized from 18, utilizing reductive amination with N-ethylpiperazine in 35% yield (Scheme 2).
Analysis of the third generation PDP 16 confirmed that it retained acceptable lipophilicity (Table 1, entry 3) but also displayed a 4.2-fold increase in affinity for recombinant pirin compared to the second generation PDP 10 ( Figure 2A) and comparable affinity for CRBN ( Figure 2B). SK-OV-3 cells were

Journal of Medicinal Chemistry
Article then treated with PDP 16 at concentrations from 50 to 1500 nM for up to 24 h ( Figure 3).
In contrast to our earlier generation PDPs, near complete pirin degradation was now observed with just 50 nM treatment and only 2 h exposure ( Figure 3A). Increasing the total initial concentration of pirin-targeting PDP 16 resulted in a clear hook effect. This bell-shaped concentration−response is consistent with the formation of a ternary complex. 22 Interestingly the hook effect was seen to decrease over time (24 h), possibly either due to the slower degradation at high concentration or from the depletion of PDP 16 due to thalidomide hydrolysis (half-life = ∼3 h, Supporting Information, Figure S1), which would reduce the effective free concentration below the negative cooperativity threshold of the ternary complex. 22 Pirin degradation was subsequently shown to be concentration-responsive with activity at concentrations as low as 0.5 nM ( Figure 3B), which was confirmed with a quantitative capillary electrophoresis-based immunoassay ( Figure 3C). The negative control benzodioxane regioisomer PDP 21 (Table 1, entry 4) displayed no pirin depletion at equimolar concentrations (Supporting Information, Figure S19), 50 and degradation was also confirmed to be proteasome-dependent by rescue following preincubation with the proteosome inhibitor, MG132 (500 nM, Supporting Information, Figure S19). 51 We then carried out whole proteome mass spectrometry to estimate the cellular selectivity of the pirintargeting PDP 16 in an unbiased manner, quantified using tandem mass tagging (TMT) ( Figure 3D, www.proteomics.com). After treating SK-OV-3 cells with 50 nM PDP 16 for 4 h, and comparing to vehicle treated cells using Benjamini−Hochberg corrected p values, we found that from 8547 quantifiable proteins identified, only pirin (2.3-fold reduction, p(adj) = 1.4 × 10 −4 ) displayed a statistically significant (p(adj) < 0.05) 52 difference in protein expression.
To confirm that our chemical probe 1 also bound pirin within SK-OV-3 cancer cells, we carried out competition experiments designed to rescue pirin depletion by PDP 16. The concentrations required for a mutually exclusive binding ligand to displace a probe molecule are dependent on the affinity of the ligand relative to the ratio of the free concentration of the probe and its affinity for the protein target complex. 53 Because the depletion of pirin is a nonequilibrium event that does not necessarily require complete target occupancy, it is more difficult to observe competition at later time points owing to continued protein turnover. 54 Pretreating SK-OV-3 cells with 10 μM thalidomide, as a CRBN-binding competitive ligand, demonstrated that after 2 h treatment with PDP 16 (5 nM), we successfully rescued pirin depletion (Supporting Information, Figure S19).

■ CONCLUSION
Exploiting cell-based phenotypic screens to identify new diseaseassociated therapeutic targets is an increasingly frequent strategy in drug discovery. While this approach can identify novel targets with unique mechanisms of action, these proteins are often poorly characterized and can lack identifiable enzymatic activity, ligands, and biomarkers of target engagement. The main focus of research into PDPs has been as potential therapeutics with novel mechanisms of action. We designed a PDP as an intracellular  (4 h) in SK-OV-3 cells compared to vehicle control, using a tandem mass tagging (TMT) MS2 protocol on the cell lysate, 8547 quantifiable proteins were identified; each blue dot represents a single quantifiable protein, pirin is marked in red (adjusted p value = 1.4 × 10 −4 ), p values were calculated using a linear modeling based t test and corrected for multiple comparisons using the Benjamini−Hochberg method to give the p(adj) values shown, dotted lines represent 2-fold depletion of the protein and a p(adj) = 0.05.

Journal of Medicinal Chemistry
Article probe against the poorly understood and noncatalytic molecular target, pirin. Developing PDPs to confirm intracellular target engagement, and potentially develop intracellular SAR, against challenging proteins, is an important addition to the current methods for compound profiling.
For PDPs to be used as target engagement probes, their rapid development and validation is crucial. The ideal strategies for efficient and successful PDP design are still under investigation and will clearly improve as more protein targets are modulated and additional crystallographic evidence of the target protein/E3 ligase/PDP ternary complexes are discovered. The number of variables involved in PDP design against nonvalidated target proteins can make the process daunting. By focusing on the physicochemical properties of our probe molecules, in only three iterations, we developed a selective degradation probe that eliminates pirin at low concentration and in a short time period. This confirmed our chemical probe 1 does bind pirin in an intracellular environment, and PDP 16 provides another chemical tool to study a largely unexplored protein.

■ EXPERIMENTAL SECTION
General Experimental. Unless otherwise stated, reactions were conducted in oven-dried glassware under an atmosphere of nitrogen or argon using anhydrous solvents. All commercially obtained reagents and solvents were used as received. Thin layer chromatography (TLC) was performed on precoated aluminum sheets of silica (60 F254 nm, Merck) and visualized using short-wave UV light. Flash column chromatography was carried out on Merck silica gel 60 (partial size 40−65 μm). Column chromatography was also performed on a Biotage SP1 or Biotage Isolera Four purification system using Biotage Flash silica cartridges (SNAP KP-Sil) or for reverse phase purifications SNAP Ultra C18 cartridges. Ion exchange chromatography was performed using acidic Isolute Flash SCX-II columns. Semipreparative HPLC was performed on an Agilent 6120 system, flow 20 mL/min, eluents 0.1% acetic acid in water and 0.1% acetic acid in methanol, gradient of 10−100% organic phase. Lipophilic method: Chromatographic separation at room temperature was carried out using a 1200 series preparative HPLC (Agilent, Santa Clara, USA) over a 15 min gradient elution (gradient 15 min, 20 mL) from 60:40 to 0:100 water:methanol (both modified with 0.1% formic acid) at a flow rate of 20 mL/min. 1 H NMR spectra were recorded on Bruker AMX500 (500 MHz) spectrometers using an internal deuterium lock. Chemical shifts are quoted in parts per million (ppm) using the following internal references: CDCl 3 (δH 7.26), MeOD (δH 3.31), and DMSO-d 6 (δH 2.50). Signal multiplicities are recorded as singlet (s), doublet (d), triplet (t), quartet (q), multiplet (m), doublet of doublets (dd), doublet of doublet of doublets (ddd), broad (br), or obscured (obs). Coupling constants, J, are measured to the nearest 0. 1 Hz. 13 C NMR spectra were recorded on Bruker AMX500 spectrometers at 126 MHz using an internal deuterium lock. Chemical shifts are quoted to 0.01 ppm, unless greater accuracy was required, using the following internal references: CDCl 3 (δC 77.0), MeOD (δC 49.0), and DMSO-d 6 (δC 39.5). High resolution mass spectra were recorded on an Agilent 1200 series HPLC and diode array detector coupled to a 6210 time-offlight mass spectrometer with dual multimode APCI/ESI source or on a Waters Acquity UPLC and diode array detector coupled to a Waters G2 QToF mass spectrometer fitted with a multimode ESI/APCI source. All compounds were >95% purity by LCMS analysis unless otherwise stated.

Journal of Medicinal Chemistry
Article (3 × 15 mL). The combined organic layer was washed with brine (20 mL) and dried (Na 2 SO 4 ) to afford the crude product as a yellow oil. This material was used directly in the next step without further purification. LCMS (ESI + ): RT = 1.57 min, 84%, (M + H) + 304.