Targeting the Deubiquitinase USP7 for Degradation with PROTACs

Abstract

Targeting deubiquitinating enzymes (DUBs) has emerged as a promising therapeutic approach in several human cancers and other diseases. DUB inhibitors are exciting pharmacological tools but often exhibit limited cellular potency. Here we report PROTACs based on a ubiquitin-specific protease 7 (USP7) inhibitor scaffold to degrade USP7. By investigating several linker and E3 ligand types, including novel cereblon recruiters, we discovered a highly selective USP7 degrader tool compound that induced apoptosis of USP7-dependent cancer cells. This work represents one of the first DUB degraders and unlocks a new drug target class for protein degradation.

Graphical Abstract

The recent advancements in therapeutic modalities enable the targeting of a plethora of proteins, which were previously deemed undruggable.¹ At present, proteolysis targeting chimeras (PROTACs) promise a paradigm shift in modern medicinal chemistry and have attracted substantial interest and investment from both academia and pharmaceutical industry.^2,3 PROTACs are bifunctional compounds comprising a target ligand, an E3 ligase binder, and a linker tethering both moieties. Instead of merely blocking the functional domain of the protein of interest (POI), PROTACs enable target degradation via hijacking of the ubiquitin-proteasome machinery.^4–6 Recently, PROTACs have shown clinical activity and favourable side effect profile in cancer highlighting the applicability as therapeutics.⁴ Deubiquitinating enzymes (DUBs) recognize and cleave the isopeptide bonds linking ubiquitin to proteins and play an important role in maintaining protein stability, homeostasis and signalling in cells.^7–9 DUBs emerged as promising targets that could, if inhibited or depleted successfully, complement the inventory of drugs for various diseases, especially cancer.^10,11 USP7 is a DUB known to be associated with a variety of tumour-suppressor genes and proto-oncogenes, pointing to its significance in tumour biology and progression.^12,13 USP7 exhibits high expression levels in numerous cancer tissues and often correlates with poor prognosis and metastasis. It is a key regulator of the tumour suppressor p53 either via direct interaction,¹⁴ or indirectly by stabilizing mouse double minute 2 homolog (MDM2), that is the E3 ligase, which facilitates ubiquitination and degradation of p53.¹⁵ Inhibition of USP7 promotes MDM2 ubiquitination and proteasomal degradation, ultimately leading to elevated levels of p53 and cell death.¹⁶ Furthermore, USP7 inhibition is active against cancer cell with TP53 loss/mutation implying that degradation of other substrates contribute to the effects of USP7 inhibitors.¹⁷ Several USP7-selective inhibitors have been developed in the last decade,^18–20 of which selected are depicted in Figure 1. ^21–23

Fig. 1.

Targeting USP7 with inhibitors and PROTACs.

The USP7 inhibitors FT671 (2) and XL188 (3) were chosen as a possible starting point for PROTAC synthesis. Notably, both scaffolds exhibited high polar surface area (Table 1), accompanied by a moderate logD value of 1–2. The (R)-configuration of the methyl or di/trifluoromethyl groups at the right-hand side tail is critical for target binding affinity of these inhibitors (Figure 2A).^21,22 The trifluoromethyl analogue (4) of XL188 was accessible via an asymmetric catalytic reduction of a prochiral cinnamic acid derivative (Scheme S1),²² and the introduction of the CF₃ group increased the lipophilicity. Unexpectedly, the USP7 inhibitory activity of 4 was lower compared to XL188 containing an (R)-CH₃ group at this position (Table 1, Figure S1). Compounds 5 and 6 with N-Boc-piperidine instead of the solvent-exposed piperazine handle (Figure 2B) showed notably decreased inhibition at 5 μM with 58% and 53% residual USP7 activities, respectively, whereas 7 inhibited USP7 with an IC₅₀ value of 1.7 ± 0.2 μM. Trends of the in vitro data were in agreement with docking results and cell viability profiles (Table 1, Figures S2–S5). Moreover, the nitrogen replacement resulted in the desired increase in logD of the ligand.

Table 1.

Overview of physicochemical properties, experimental and computational USP7 binding data, and impact on cell viability of USP7 inhibitors 2–7.

cmpd	TPSAa (Å2)	logDb	PPBc (%)	IC50d (μM) or RA (%)	MM-GBSA ΔG Binde (kcal/mol)	cell viabilityf (%)
2	109	2.5	90	0.032 ± 0.007	−101.8	41
3	109	1.3	87	0.31 ± 0.51	−98.4	43
4	109	1.8	88	0.95 ± 0.20	−101.7	n.d.g
5	132	3.6	96	58 ± 9%	−91.9	61
6	132	3.1	n.d.	53 ± 10%	−92.1h	52
7	132	3.2	94	1.70 ± 0.21	−94.7	48

^aTopological polar surface area is given in Ų.

^bExperimental partition coefficient.

^cPlasma protein binding (PPB) values were estimated by an HPLC-based method.

^dInhibition data obtained with isolated USP7 catalytic domain using Ub-AMC as a substrate. Residual enzyme activities (RA) at 5 μM of compounds 5–6 are given. The IC₅₀ value for FT671 is in accordance with literature data,²³ whereas in our assay, XL188 showed marginally lower USP7 inhibition than reported previously.²¹

^eAn estimate for the binding affinity derived from computational docking (see Supporting Information). A more negative value indicates stronger binding.

^fRemaining cell viability after 24 h treatment of MM.1S cells with 10 μM of each compound.

^gnot determined.

^hMean of (R)- and (S)-isomers.

Fig. 2.

(A) The crystal structure of USP7 in complex with XL188 (PDB: 5VS6)²¹ reveals a possible exit vector for linker attachment; (B) chemical structures of modified USP7 inhibitors along with candidates suitable for linker attachment.

To enable quick access to PROTACs, we applied the racemic compound 6 to screen different linker lengths of PROTAC molecules. As both incorporation of a von Hippel-Lindau or an inhibitor of apoptosis protein E3 ligand would result in USP7 PROTACs with unfavourable high TPSA,²⁴ we chose cereblon (CRBN) as E3 ligase and synthesized a series of eight PROTACs that are listed in Table S1. From this initial series, the rigidified PROTAC 9 performed best in decreasing USP7 levels in MM.1S cells at a concentration of 1 μM (Table 2 and Figure S6). Subtle structural modifications of this hit compound revealed structure-degradation relationships. Replacement of the racemic CH₃ group with an (R)-CF₃ substituent in 16 increased both degradation potency and lipophilicity, but PROTAC 17 with an (R)-CH₃ group performed best, which is also reflected by a prominent USP7 inhibition in vitro with an IC₅₀ value of 1.6 ± 0.3 μM (Figure S1). Accordingly, further compounds with this group were synthesized. In 18 and 19, one or two carbon atoms in the linker chain were removed, thus increasing their rigidity by eliminating rotatable bonds. However, such modifications negatively affected the degradation potency of new PROTACs. PROTAC 20 possesses a piperazine exit vector whose nitrogen atoms might be protonated under physiological conditions. Surprisingly, the introduction of an aromatic fluorine atom at the CRBN-binding moiety (21 and 22) was detrimental. To reduce the overall size of the degrader, alternative CRBN binders were incorporated in 23 and 24. Notably, such binding motifs were included in recent patent literature,²⁵ but their syntheses have not yet been described in peer-reviewed journals. Compounds derived from anilinic CRBN binders C and D were less active compared to PROTAC 17. Compound 25 (Table 2) represents a CRBN non-binding negative control, as methylation of the imide nitrogen abolishes E3 engagement.²⁶ As expected, no USP7 degradation was observed with this compound. Physicochemical properties of all rigidified PROTACs and western blots are provided in Tables S2 and Figure S6. We observed a narrow activity window among highly related rigidified compounds with comparable physicochemical features. These findings suggest a high dependency on productive USP7 : PROTAC : E3 complexes. Out of the entire series of 18 novel PROTACs, 17 showed the most pronounced USP7 depletion. Its synthesis is outlined in Scheme S1.

Table 2.

Overview on structures, physicochemical properties, and cellular activities of rigidified USP7-targeting PROTACs.

^aTopological polar surface area is given in Ų.

^bExperimental partition coefficient.

^cUSP7 depletion after 24 h treatment of MM.1S cells with 1 μM of each compound.

Due to the striking USP7 degradation capacity of 17 (CST967), this drug was subjected to extensive biological studies. The degradation profile of the USP7 PROTAC 17 was determined by western blot analysis performed in the multiple myeloma cell line MM.1S. Increasing the PROTAC concentration enhanced the rate of USP7 degradation, eventually reaching a slight hook effect at 10 μM. Maximal degradation (D_max) of 85% at a concentration of 1 μM and DC_50/24h potency of 17 nM was observed. In contrast, no effect on the IMiD-induced CRBN neosubstrate GSPT1 and only slight degradation of IKZF1 was seen (Figure 3). Given the dynamic nature of degradation, a time-course analysis from 3 to 96 hours was performed in MM.1S cells to assess the degradation response over time. We observed maximal degradation at 24 hours which was followed by the gradual recovery of USP7 levels (Figure S7). Furthermore, we noted upregulation of the direct USP7-target p53 and induction of cleaved-PARP, a hallmark of cell death. To discern USP7 protein turnover and cell recovery, a drug washout experiment was performed after 24 hours. USP7 levels reverted 72 hours after recovery. In congruence with this, upregulation of p53 and cleaved-PARP levels were observed (Figure S8). PROTAC-mediated USP7 degradation was abrogated by the proteasome inhibitor MG132 and neddylation activation enzyme (NAE) inhibitor MLN4924 confirming degradation via the ubiquitin-proteasome pathway. Similarly, USP7 degradation was blocked by 100-fold excess pomalidomide, which competes for binding with CRBN (Figure S9), and in a genetically modified MM.1S cell line harbouring CRBN knockout, no USP7 depletion was observed, both corroborating USP7 degradation via the E3 ligase CRBN (Figure S10). A global-proteome analysis of the downstream effects of the USP7 PROTAC 17 was performed using diaPASEF-based mass spectrometry²⁷ in MM.1S upon treatment with the PROTAC at 0.1 μM for three hours. Of the total of 7170 unique proteins identified (Figure 4A), USP7 appeared as the most significantly downregulated protein, thus further confirming the degradation ability of the USP7 PROTAC. The only other significantly regulated protein was Cox5A. No changes for known IMiD-induced CRBN (neo-)substrates were observed. Of note, neither in western blots nor in the proteomic data we observed changes of CRBN protein levels, arguing against CRBN deubiquitination and stabilization via USP7. To further explore the degrading efficacy and potency of the USP7 PROTAC 17, we tested a panel of cancer cell lines that are known to be dependent on USP7. Western blot analysis confirmed degradation of USP7 to varying degrees. The USP7 PROTAC led to decreased cell viability in all cell lines, thus underlining the anti-cancer activity of the compound (Figures 4B, S11, S12).

Fig. 3.

PROTAC 17 induces degradation of USP7. Western blot analysis of USP7, IKZF1, GSPT1 and CRBN protein levels in MM.1S cells treated with compounds for 24 h (left). DC₅₀ values were obtained by fitting D_max values to a variable slope dose−response model (middle). CellTiter-Glo luminescent cell viability assay (right) showing effect of USP7 PROTAC upon 96 h treatment. Data are shown as mean±s.d. (n=2,3 respectively).

Fig. 4.

(A) diaPASEF-based mass spectrometry was performed in MM.1S cells upon treatment with USP7 PROTAC 17 at a concentration of 0.1 μM for 3 h. A total of 7171 identified proteins were plotted as log2 fold change (USP7 PROTAC/DMSO) versus–log10 of p-value with a threshold of 1% FDR; (B) western blots showing degradation of USP7 in A549 or LNCAP cell lines upon treatment with USP7 PROTAC 17 for 24 h (above). CellTiter-Glo luminescent cell viability assay upon a 96 h treatment with USP7 PROTAC 17 in the respective cell lines (below). Data are shown as mean±s.d. (n=3).

In our proof-of-concept study, we describe one of the first DUB degraders. Our lead PROTAC 17 is highly potent and selective for USP7 and consistently shows activity in USP7-dependent cancer cells. Quantitative proteomics confirmed its target selectivity. We propose that such a tool will be invaluable to unravel novel substrates and physiological functions of USP7. While the early generation of USP7 inhibitors only partially and temporally modulates protein function, our degrader leads to an entire loss of the protein,²⁸ which could unveil the full potential of USP7 targeted therapies.