Introduction

Interactions between E3 ubiquitin ligases and their substrates occur through “degrons” - short linear motifs (SLiMs) typically located within intrinsically disordered regions (IDRs) of substrates and which have relatively weak, micromolar affinities for the E3 (Min, Mayor, and Lindon 2013; Guharoy et al. 2016). Degradation of a substrate requires extensive sequential ubiquitination of the substrate through multiple rounds of recruitment of a ubiquitin-loaded conjugating E2 enzyme to the E3-substrate complex, and proteasome recognition depends on the number, type and length of polyubiquitin chains. In many examples explored to date, target discrimination and productive complex formation (i.e., leading to degradation of substrate) relies on the interaction of the E3 with multiple degrons ((Karamysheva et al. 2009; Fiore et al. 2016; Pierce et al. 2016; Tian et al. 2012; Okoye et al. 2022), and there is often a correlation between E3-substrate affinity and rate and/or timing of degradation.

Progression of cells out of mitosis involves ordered ubiquitin-mediated destruction of at least 100 different protein targets under control of a large multi-subunit E3 ubiquitin ligase, the anaphase promoting complex/cyclosome (APC/C), bound to one of its two coactivators Cdc20 and Cdh1/FZR1 (Davey and Morgan 2016; Min, Mayor, and Lindon 2013; Bakos et al. 2018). These activators contain a WD40 domain that binds substrate degrons flanked by partially disordered regions that mediate binding to neighbouring APC/C subunits, resulting in conformational change that enhances binding of the E2. APC/C^Cdc20 is a key component of the cell cycle machinery, with full activation of APC/C^Cdc20 acting as the trigger for mitotic exit through targeted degradation of mitotic cyclins and securin (PTTG1) (Meadows and Millar 2015). Coordination of mitotic exit events with segregation of duplicated chromosomes requires careful control of APC/C^Cdc20 activity, which is achieved via the mitotic checkpoint that inhibits APC/C in the presence of faulty chromosome attachments to the mitotic spindle (Izawa and Pines 2011; Hein and Nilsson 2014; Fiore et al. 2016; Qiao et al. 2016; Alfieri, Zhang, and Barford 2017; Watson et al. 2019). The mitotic checkpoint complex (MCC) prevents APC/C^Cdc20 targeting of critical metaphase substrates, whilst allowing degradation of a small number of so-called “checkpoint-independent” substrates such as cyclin A2 and Nek2A (Geley et al. 2001; Hayes et al. 2006). As cells exit mitosis, APC/C^Cdc20 activity declines and is replaced with APC/C^FZR1, which maintains activity until the end of G1 phase.

In the past 15 years, high-resolution X-ray and cryo-EM (electron microscopy) studies of the APC/C and its interactions with substrates and E2s has generated a detailed description of the structure-function relationships that drive ubiquitination and degradation (Barford 2020). The binding of Cdc20 or FZR1 to the core APC/C creates at least three degron-binding sites for the known APC/C degrons, namely the “Destruction-box” (D-box, consensus RxxLxxxxN) and KEN motifs, and the ABBA motif thought to be required for Cyclin A degradation only(Qin et al. 2017). A cryo-EM study of the structure of APC/C-FZR1 in complex with its pseudo-substrate inhibitor Acm1 revealed simultaneous engagement of D-box, KEN, and ABBA motifs of Acm1 with their respective receptor sites on the interactions (He et al. 2013). The KEN motif docks to the top surface of the WD40 propeller of the co-activator and the D-box to a cleft formed between two blades of the propeller and the neighbouring APC10 subunit such that substrate engagement with degron receptors is likely to stabilize the active complex (Burton and Tsakraklides 2005; Buschhorn et al. 2010; Da Fonseca et al. 2010; L. Chang et al. 2014; Matyskiela and Morgan 2009; Qin et al. 2019). The critical residue of the D-box, leucine at position 4 (P4), contacts a hydrophobic pocket in the co-activator subunit, and the ‘tail’ of the D-box degron and its flanking sequence (P8-12) contact the APC10 subunit. Mechanisms by which the APC/C “orders” the degradation of its substrates include co-activator switching, fine-tuning of APC/C-substrate interactions by phosphorylation or other post-translational modifications, differential processivity of ubiquitination, and substrate competition, in addition to differential degron-binding affinities (Davey and Morgan 2016; Alfieri, Zhang, and Barford 2017; Bodrug et al. 2021; Okoye et al. 2022). Most recently, single-molecule studies have shed new insights into the key role of degron multivalency in enabling efficient substrate ubiquitination and degradation (Hartooni et al. 2022).

Inhibitors of APC/C^Cdc20 activity represent an interesting therapeutic approach to target dividing cells in cancer. Given the large size of the APC/C machine (11 subunits) and the complex mechanisms described above that regulate its function, it is not surprising that it is challenging to target. Apcin and TAME are recently identified small-molecule inhibitors, but they have limited activity and complicated output (Richeson et al. 2020; Sackton et al. 2014a). In this paper we use a rational approach, based on D-box consensus sequences and a ‘Super D-box’ peptide derived from Hsl1, and examination of the Cdc20-degron interface, to design a series of more potent binders containing unnatural amino acids aimed at better filling the hydrophobic pocket on the interaction interface. We quantified binding by thermal shift assays (TSA) and surface plasmon resonance (SPR) and used a cellular thermal shift assay (CETSA) to demonstrate target engagement within the cellular context. The peptides also show functional engagement with APC/C in the cell as evidenced by their ability to drive the degradation of a fluorescent protein. Most strikingly, in vitro ubiquitination assays with recombinant APC/C^Cdc20 shows that these peptides are more potent inhibitors of Cyclin B1 ubiquitination than Apcin. Interestingly, we find that although inhibitory activity of the peptides correlates with Cdc20-binding affinity, their degradation efficacy does not. This may be due to the complex nature of APC/C degrons and their bipartite interaction with different subunit, role in E2 recruitment, and consequent impact of positioning for effective ubiquitination. The results are a useful starting point for the further development of these peptides as molecular therapeutics for blocking APC/C as well as potentially also for harnessing APC/C for targeted protein degradation.

Materials and Methods

Cloning, expression, and purification of Cdc20

DNA encoding residues 161-477 of human Cdc20 (Cdc20^WD40) was cloned into a pU1 vector with an N-terminal His₆-tag followed by a TEV protease cleavage site. Plasmid was then transformed into DH10 MultiBac cells expressing a Cre-recombinase. Positive clones were grown up and bacmid DNA prepared by standard protocols. Sf21 or Sf9 cells were grown at 27°C in Erlenmeyer flasks (Corning) and maintained in mid-log phase of growth prior to all experiments. High titre baculovirus was produced by transfecting bacmid DNA into Sf21 cells at 0.5 × 10⁶ cells/ ml cells using Superfect (Qiagen) in 24 deep-well blocks. Virus was harvested 1-week post transfection. For protein over-expression, Sf21 or Sf9 cells were infected with the virus stock and harvested about 60 hours post infection. Cell pellets were resuspended in 50 mM Tris-HCl, 300 mM NaCl, 1 mM MgCl₂, 1 mM TCEP, 5% (v/v) glycerol, SigmaFAST EDTA-free protease inhibitor cocktail (1 tablet/100 ml), pH 8.5. Resuspended pellets were lysed by one freeze-thaw cycle at -80°C. Lysates were then clarified by centrifugation at 45,000 × g for 45 minutes at 4°C. Supernatants were flowed over a 5 ml HisTrap Excel column and washed with 20 column volumes (CV) of 50 mM Tris-HCl, 300 mM NaCl, 10 mM imidazole, 1 mM MgCl₂, 1 mM TCEP, 5% (v/v) glycerol, pH 8.5. Proteins were then eluted with the above buffer including 300 mM imidazole directly into a 26/10 desalting column pre-equilibrated in the above buffer without imidazole. Eluted protein fractions were then pooled, and the His₆-tag was removed using His₆-TEV protease (S219V) overnight at 4°C. Proteins were then flowed back over a 5 ml HisTrap Excel column, collecting the flow-through containing the non-tagged Cdc20 protein. Protein eluent was then diluted in 25 mM Tris-HCl buffer, 1 mM MgCl₂, 1 mM TCEP, 5% (v/v) glycerol, pH 8.5 to a final concentration of 30 mM NaCl. Proteins were then loaded onto a MonoQ 10/100 GL column and eluted over 20 CV with 1 M NaCl. Protein fractions containing the Cdc20 protein were then pooled and concentrated before separating on a Superdex 75 increase 10/300 GL column in the final buffer containing 25 mM Tris-HCl, 150 mM NaCl, 1 mM MgCl₂, 5 mM TCEP, 5% (v/v) glycerol, pH 8.5.

Minimal biotinylation of Cdc20 for SPR

The protocol for minimal biotinylation was adapted from Papalia and Myszka (Papalia and Myszka 2010). Purified Cdc20^WD40 was diluted into the reaction buffer (10 mM Tris-HCl, 150 mM NaCl, 1 mM MgCl₂, 5% (v/v) glycerol, 1 mM TCEP, pH 7.3). A 0.9:1 molar ratio of Sulfo-NHS-LC-LC-Biotin (Thermo Fisher Scientific, A35358) was added to the diluted Cdc20 protein. The contents were briefly mixed by vortex and incubated on ice for 3 hours. The sample was then separated on a Superdex 75 Increase 10/300 GL column to remove free biotin.

Peptide synthesis and purification

Peptides synthesis was performed on a 0.1 mmol scale using Ramage-ChemMatrix® resin (Sigma Aldrich). Fmoc-L-amino acids (2 eq.), HATU (2 eq.) and HOAt (2 eq.) were dissolved in 2 mL of NMP. DIEA (3.4 eq.) were used to activate the coupling mixture. Activated Fmoc-L-amino acids were coupled for 10 minutes (Fmoc-L-Arginine, 2 × 5 eq., 30 minutes). Resins were washed in DMF and deprotected in 20% piperidine in DMF for 15 minutes. All peptides were N-terminally acetylated in 4 ml DMF, 4 mL acetic anhydride, 2 mL DIEA for 10 minutes. A peptide cleavage cocktail consisting of 93% TFA, 3.5% TIPS and 3.5% ddH₂O was used to deprotect and cleave the peptide from the resin for 1 hour. The eluate was triturated by the addition of diethyl ether and the resulting precipitate was isolated by brief centrifugation. All peptides were characterised by LCMS using a Waters ACQUITY H-Class UPLC with an ESCi Multi-Mode Ionisation Waters SQ Detector 2 spectrometer. LC was performed on a ACQUITY UPLC® CSH C18 (2.1 mm × 50 mm, 1.7 µm, 130 Å) at 40°C, with a PDA eλ detector 220 – 800 nm, interval 1.2 nm. The following solvents and gradients were used for LC runs. Solvent A: 2 mM NH₄OAc in 95% H₂O, 5% MeCN, solvent B: 100% MeCN, solvent C: 2% Formic acid from 5-95% B with a constant of 5% C over 1 minute at 0.6 ml/min. Analytical and semi preparative HPLC runs were performed on an Agilent 1260 Infinity system using a Supelcosil ABZ+PLUS (150 mm × 4.6 mm, 3 µm) and Supelcosil ABZ+PLUS (250 mm × 21.2 mm, 5 µm), respectively. Peptides were eluted with a linear gradient system (solvent A: 0.1% TFA in H₂O, solvent B: 0.05% TFA in MeCN) over 15 minutes at 1ml/min and 20 minutes at 20ml/min, respectively. Eluents were monitored by UV absorbance at 220 nm and 254 nm. Analytical data for all peptides are shown in Figure S5.

Thermal-shift assays (TSA)

Assays were performed using a Roche LightCycler 480 I in 96-well plate format. Each well (20 µl) was prepared with 750 nM of purified Cdc20^WD40 and varying concentrations of D-box peptides, Apcin or DMSO (vehicle control) in assay buffer; 25 mM Tris-HCl, 150 mM NaCl, 1 mM MgCl₂, 5% (v/v) glycerol, 1 mM TCEP, 1% (v/v) DMSO, 5x SYPRO Orange (Thermo Fisher), pH 8.5. Thermal ramps were conducted from 25°C to 95°C at a rate of 0.03°C/sec and data were collected at a frequency of 20 points/°C. An excitation wavelength of 483 ± 35 nm was used to excite SYPRO Orange, and the fluorescence emission was detected at a 568 ± 20 nm. Measurements were performed in triplicate and errors listed are the standard deviation. Melting temperatures were determined by the minima peak of the negative differential in the ‘Tm calling’ analysis within the in-built analysis software.

Surface Plasmon Resonance (SPR) assays

Experiments were performed using a Biacore T200 instrument (GE healthcare) at 15°C. Biotinylated-Cdc20^WD40 was immobilised onto a SA biosensor chip (GE healthcare) in running buffer (10 mM HEPES, 150 mM NaCl, 0.1 mM TCEP, 0.05% (v/v) Tween 20 and 1% (v/v) DMSO, pH 7.4) over flow cells 2, 3 and 4 at varying ligand densities. Flow cell 1 was used as a reference cell. Free biotin binding sites were blocked using amine-PEG₄-Biotin. Peptides and Apcin analytes were diluted from DMSO stock solutions in running buffer without DMSO and were buffer matched to 1% DMSO. Titrations of each analyte were run over the sensor chip at a flow rate of 30 µl/min. Binding interactions were detected as a change in response units over the reference flow cell and subtracted from a blank buffer injection. Dissociation constants (K_D) were calculated by fitting the response units (RU) at steady-state equilibrium generated by the binding of an analyte to Cdc20^WD40 against the concentration of analyte using the following equation:

where RU_analyte is the response units at equilibrium during a given injection of a concentration of analyte, [analyte]. RU_max is the maximum response produced by the a given analyte, dependant on the RU of immobilised ligand on a given flow cell. K_D is the dissociation constant of a given analyte to the ligand. K_D values are shown as the average of measurements from the three reference-subtracted flow cells.

Cellular thermal shift assays (CETSA)

Full-length Cdc20 (residues 1-499) with a C-terminal HiBiT tag (GSVSGWRLFKKISGS, Promega) was cloned into a pcDNA3.1(-) vector. HEK 293T cells were cultured in DMEM + 10% FBS (Sigma Aldrich, F7524) at 5% CO₂ in a humidified environment. Cells were grown to 70% confluency in T75 flasks prior to transient transfection with 10 µg of Cdc20_HiBiT_pcDNA3.1(-) plasmid with Lipofectamine 2000 (Invitrogen, ThemoFisher Scientific) according to the manufacturers’ protocol. Cells were harvested after 48 hours by trypsinisation and were subsequently washed twice in PBS with repeated centrifugation at 1000 × g for 2 minutes. The pellet was then resuspended in lysis buffer (PBS, 1 × SigmaFAST EDTA-free protease inhibitor tablet (Sigma Aldrich), 2 mM NaVO₃, 5 mM NaF, pH 7.4) and freeze-thaw lysed in liquid nitrogen. The lysate was clarified by centrifugation at 20,000 × g, 4°C for 20 minutes and the protein concentration of the supernatant was quantified by BCA (Pierce). Lysates were used at a final concentration of 0.2 mg/ml in lysis buffer. Lysates were aliquoted in 300 µL and were spiked with D-box binding site ligands to a concentration of 100 µM maintaining 1% DMSO. Compounds were incubated on ice for 30 minutes prior to aliquoting further into PCR strip tubes on a PCR block at 4°C. Lysate aliquots were then heated on a second PCR block at the indicated temperatures for 3 minutes prior to returning to 4°C. Heated lysates (5 µl) were then transferred into an AlphaPlate light-grey 384-well plate in quadruplicate by multichannel pipette. Nano-Glo HiBiT lytic detection system (Promega) was diluted as per the manufacturers’ instructions and 5 µl were added to each well by multichannel pipette. Lysis buffer and a non-transfected HEK 293T cell lysate were used as negative controls. Following five minutes of incubation on a plate shaker, the plate was measured using a CLARIOStar microplate reader (BMG Labtech), with the detector set to read at 460 ± 80 nm, the focal height at 10.5 cm and the gain adjusted to 2000. Data were normalised to the unheated sample (4°C) and were fitted using a Boltzmann equation to extract the melting temperature (T_m) (Niesen, Berglund, and Vedadi 2007).

Protein crystallisation

Peptide D21 was added to Cdc20^WD40 in a stoichiometric manner and was co-concentrated to 1.9 mg/ml. The resulting complex was crystallised in a 2:1 protein to well solution ratio at 20°C using the sitting-drop vapour-diffusion method with a well solution containing 0.1 M MES pH 6.5, 12% (w/v) PEG 6000, 10% (v/v) MPD for Cdc20^WD40-D21 and 0.1 M MES pH 6.5, 14% (w/v) PEG 6000, 10% (v/v) MPD for Cdc20^WD40-D20 and Cdc20^WD40-D7. Crystals grew to a maximum size after 3 days of incubation. For soaking experiments, crystals were first looped and washed through three drops containing 0.1 M MES pH 6.5, 20% (w/v) PEG 6000 to wash out MPD from the crystal. Crystals were then looped and incubated in a solution containing 0.1 M MES pH 6.5, 20% (w/v) PEG 6000 and 2.5 mM D21 or D20 (5% (v/v) DMSO) or D7 (10% (v/v) DMSO) for four hours. Soaked crystals were cryo-protected in the soak solution supplemented with 10% (v/v) glycerol and were flash frozen in liquid nitrogen.

Data collection and structure determination

Diffraction data were collected on beamline I04 at the Diamond Light Source (Oxford, UK) and processed using autoPROC-STARANISO STARANISO (Vonrhein et al. 2018). Phases were obtained by molecular replacement using the crystal structure of human Cdc20 (PDB ID code 4GGC) as the search model (Tian et al. 2012). Iterative model building and refinements were performed with COOT and BUSTER, respectively (Emsley 2010, Bricogne G. et al) Cdc20-D20 and Cdc20-D21 datasets were first refined using Refmac5 within the CCP4i suite ((Winn et al. 2011; Kovalevskiy et al. 2018) before final refinements using BUSTER. Data collection and structure refinement statistics are summarised in Table S1.

Ubiquitination assays

In vitro ubiquitination experiments were performed using APC/C and Cdc20 purified from insect cells (Zhang et al. 2016). 60 nM APC/C, 30 nM Cdc20, 90 nM UBA1, 300 nM UbcH10, 300 nM Ube2S, 35 mM ubiquitin, 1 mM cyclin B1, 5 mM ATP, 10 mM MgCl₂, were mixed in a buffer containing 40 mM HEPES (pH 8.0), 80 mM NaCl, 0.6 mM DTT. The reaction was either performed with the indicated concentrations of peptides or DMSO (Sigma Aldrich) as the vehicle control. The reaction was incubated for 30 min at 23 °C and stopped by the addition of one volume of 2x concentrated NuPAGE LDS loading buffer (Invitrogen).

Protein degradation assays

The pEGFP-N1 vector was modified by swapping the EGFP-coding sequence for mNeon-coding sequence using the AgeI/NotI cloning sites. The Aurora kinase A (AURKA) C-terminal fragment (364-403) containing the non-degron R₃₇₁xxL motif (D0) together with an extended IDR was amplified by PCR and cloned into the modified vector with BamHI/AgeI sites. Round the horn site-directed mutagenesis was used to generate different D-box variants and validated by DNA sequencing. U2OS cells were cultured in DMEM supplemented with 10% FBS, 200 µM Glutamax-1, 100 U/ml penicillin, 100 µg/ml streptomycin, and 250 ng/ml fungizone (all from ThermoFisher Scientific) at 37ºC in humidified atmosphere containing 5% CO₂. Plasmids were introduced into U2OS cells by electroporation using the Neon™ Transfection System 10 μL Kit (ThermoFisher Scientific) and cells seeded on eight-well microscopy slides (Ibidi) and recovered for 24 hours. DMEM medium was exchanged for phenol red-free Leibovitz’s L15 (ThermoFisher Scientific), supplemented as above. Time-lapse imaging was conducted at 37°C using a widefield imaging platform composed of Olympus IX83 motorized inverted microscope, Spectra-X multi-channel LED widefield illuminator (Lumencor, Beaverton, OR, USA), Optospin filter wheel (Cairn Research, Faversham, UK), CoolSnap MYO CCD camera (Photometrics, Tucson, AZ, USA), automated XY stage (ASI, Eugene, OR, USA) and climate chamber (Digital Pixel, Brighton, UK), all controlled using Micro-Manager software (Edelstein et al. 2014). Fluorescence and phase contrast images of cells in mitosis were acquired with a 40X objective binned at 2x2 at 2-minute intervals. Fluorescence intensity of mNeon in individual mitotic cells was quantified from 16-bit tiff files using ImageJ, by integrating pixel measurements after subtraction of background fluorescence. Degradation curves were synchronized in silico to anaphase onset to generate average curves for multiple cells in each experimental condition.

Results

Quantification of Cdc20-binding activity of the small molecule Apcin

We first produced Cdc20 protein in sufficient quantities for biophysical analysis and then used the known small molecule binder, Apcin, to test that the purified protein was functional and to benchmark our peptide-binding measurements. As Cdc20 comprises a WD40 domain that binds to the different degrons and is flanked on each end by long intrinsically disordered regions, we made a construct comprising the WD40 domain (residues 161 to 477) with an N-terminal His₆-tag and expressed this protein in baculovirus as previously described (Sackton et al. 2014b; Tian et al. 2012). We biotinylated Cdc20 at a single-site, as shown by electrospray-ionisation mass spectrometry (Fig. S1). Using TSA and SPR, we confirmed that the purified Cdc20 was capable of binding to Apcin. The K_D obtained by SPR was 420 ± 50 nM (Fig. 2).

Figure 1.

Figure 2.

Design of D-box peptides

Structures of the D-box-APC/C interactions [(Chao et al. 2012; He et al. 2013)], which show that there are three key residues, Arginine 1, Leucine 4 and Asparagine 9 of the RxxLxxxxN motif (Fig. 3A). As a starting point we used two peptides, a 10-residue consensus-like sequence derived from Hsl1 (D1: GRAALSDITN) (Burton and Tsakraklides 2005; Frye et al. 2013; Davey and Morgan 2016), and a 9-residue consensus sequence based on known D-box degrons from APC/C substrates (D2: RLPLGDISN) (He et al. 2013). TSA and SPR showed that D1 binds to Cdc20^WD40 with a weak affinity (K_D = 18.6 ± 0.2 µM) (Table 1). D2 had no detectable affinity by TSA and was consequently not analysed by SPR. We hypothesised that the apparent lack of binding may be due the low solubility in aqueous buffer of D2 rather than an inability to bind. Based on the Cdh1-Acm1 structure, the sidechain of the amino acid at position 2 is likely to be solvent exposed in the context of Cdc20^WD40 (Fig. 3A). A substitution was therefore made at this position, from Leu to Ala, which improved peptide solubility and was therefore incorporated in all subsequent peptides.

Table 1.

Figure 3.

Isoleucine at position 7 and Proline at position 3 of the D-box peptide are optimal for binding

From the consensus sequence, we observed that substrate proteins have approximately equal frequency of Val, Leu and Ile at position 7. Based on the yeast Cdh1-Acm1 X-ray crystal structure (Fig. 3A), this interaction appears atypical of hydrophobic interactions given the largely solvent-exposed nature of the amino acid sidechain. Given the similar structural and physical properties of the three aliphatic sidechains, we compared peptides with each of these three amino acids at position 7 and found that D4 with Ile7 had the highest affinity for Cdc20 (1.5-fold higher than D5 with Leu7 (19.6 ± 0.2 µM and 27 ± 1 µM respectively) (Table 1 and Fig. S2C, D). Interestingly, the shorter hydrocarbon chain of Val in D3 gave the weakest affinity, with a K_D determined by SPR at 54.4 ± 0.7 µM.

We next investigated the contribution of Proline versus Alanine at position 3 (Table 1). Like the position 7 residues, Pro and Ala appear in approximately equal distribution to each other among known substrate proteins. In the context of D-box degron binding, modelling of our D4 peptide to the S. cerevisiae Cdh1 structure showed that Pro 3 may form a favourable turn in the D-box peptide backbone to allow the side chain of Leu 4 to adopt its canonical pocket (Fig. 3A). To test this hypothesis, we proceeded to synthesise D10, a derivative of D4 containing an A3P single point mutation. As expected, this mutation was significantly detrimental with an affinity of 70 ± 3 µM by SPR and in parallel a loss of thermal stabilisation by TSA (Fig. 3B). Upon confirming our hypothesis, we synthesised a derivative of D1 containing the A3P point mutation, yielding D19 (RAPLSDITN). This substitution resulted in 3-fold increase in affinity (K_D = 5.9 ± 0.1 µM) compared with its parental sequence (Fig. S2F, Table 1).

Unnatural amino acids at position 4 of the D-box peptide result in significantly enhanced binding affinity to Cdc20

The surface topology of Cdc20 is largely flat, making it hard to drug. Nevertheless, in Apcin the tri-chlorinated moiety makes particular use of the Leu 4-binding pocket on Cdc20. Taking inspiration from the small molecule, we explored candidate unnatural amino acids to incorporate into the D-box peptides at position 4. Given that the pocket can accommodate a tri-chlorinated carbon moiety within Apcin, we explored similar moieties to append to our D-box peptides. We incorporated (S)-2-amino-4,4-dimethylpentanoic acid (C₃) (Fig. 4A) into the backbone sequences of D4, D10 and D19 replacing Leu at position 4, yielding peptides D7, D12, and D20, respectively (Table 2). As expected, the structure-activity relationship (SAR) held true between all peptides, whereby incorporation of the unnatural amino acid increased the binding affinity over 6-fold versus the respective parental peptide (Table 2). Building on this success, we further explored the commercially available halogenated analog, (S)-2-amino-4,4,4-trifluorobutanoic acid (F₃) (Fig. 4A), leading to peptide D21 (Table 2). With the tri-fluorinated group, a further increase in binding affinity was achieved (K_D = 520 ± 10 nM), which is similar to that of Apcin.

Table 2.

Figure 4.

Crystal structures of Cdc20-peptide complexes reveal D-box binding mode

Previous attempts to co-crystallise Cdc20 and securin-derived or cyclin B1-derived D-box peptides by Tian and co-workers were unsuccessful (Tian et al. 2012), which may be due to the low affinity of peptides comprising these sequences. Despite the relatively high affinity of D21 and the approximate 1:1.5 ratio of protein to peptide used in co-crystallisation experiments, crystals were absent of peptide ligands and instead contained the 2-methyl-2,4-pentanediol (MPD) molecule in the Leucine-binding cleft (data not shown), originating from the crystallisation well solution. We therefore adopted a similar protocol to that described by Sackton et al., whereby MPD was ‘washed’ out from the crystal prior to performing a soaking experiment with the desired ligand. We attempted these soaking experiments with our four highest affinity peptides, D21, D20, D7 and D19 (in order of highest affinity to lowest) and were able to observe sufficient ligand density for all but D19.

The crystal structures of Cdc20^WD40 in complex with each of the other three D-box peptides (Fig. 5A-C) show that they bound to Cdc20^WD40 at the canonical D-box degron binding site, with a largely similar topology to the S. cerevisiae Acm1-Cdh1 structure (Fig. 5D (overlay of D21 with Acm1 D-box). The R1 guanidino group of peptides interacts forms hydrogen bonds with the carboxylic acid side chains of D177 and E465 of Cdc20^WD40. The nitrogen backbone atom of the (S)-2-amino-4,4,4-trifluorobutanoic acid/ (S)-2-amino-4,4-dimethylpentanoic acid unnatural amino acids also form a hydrogen bond with the carbonyl of D177. Additionally, the carbonyl of S5 belonging to D21/D20 form a H-bond with the nitrogen backbone atom of D177. Lastly, D6 forms inter-molecular H-bonds with R174. We also observed intra-molecular H-bond between the carbonyl of A2 with the amine of G5/S5, in addition the carbonyl of A2 to the hydroxyl of S5 in D21/D20. Crystal packing of an adjacent asymmetric unit of the WD40 domain likely occludes the assumed binding site for the C-terminal three residues (…ITN-NH₂). We therefore presume this is the reason for lack of observed density in this region of the peptides D20 and D21 (Fig. S3E and S3F, respectively). We extend this observation to further explain why we were unable to observe peptide density for the D19-soaked crystals. Specifically, our affinity data highlights a role of position 7 in binding, which in crystallo is unable to be realised. This hypothesis also correlates to the comments made by Tian et al. in their attempt to co-crystallize securin D-box peptides with Cdc20, in the identical space group (Tian et al. 2012).

Figure 5.

D-box peptides bind to Cdc20 in the cellular context

We next investigated whether the four highest-affinity peptides D21, D20, D7, and D19 can bind to Cdc20 in the cellular context using a cellular thermal shift assay (CETSA) (Martinez Molina 2013). Sackton et. al previously demonstrated that Apcin can stabilise endogenous Cdc20 by using an isothermal CETSA method (Sackton et al. 2014b). We were able to reproduce this ligand-induced stabilisation of Cdc20 using the more commonly used temperature gradient approach by densitometric analysis of western blots (Fig S3A). However, due to the low-throughput of the assay we also explored a more high-throughput approach by making use of Promega’s split-luciferase HiBiT tag appended to the C-terminus of full-length Cdc20, based on protocols previously described by Martinez and co-workers (Martinez et al. 2018). Notably, the signal is more sensitive and has a larger range of compared to a western blot, and it removes a significant time-consuming centrifugation step from the workflow. We first confirmed that omitting the centrifugation step did not significantly affect the observed T_m of vehicle control samples (Fig. S4B). To further validate that the transfected Cdc20 is functional, we probed binding of 100 µM Apcin, which gave a T_m of 54.4°C ± 0.6 °C (Fig. S4C). We then explored whether the D-box peptides at a fixed concentration stabilise the Cdc20, and for D7, D20 and D21 we observed increases in the thermal stability of Cdc20 that correlated with their binding affinities as previously determined (Fig. 6 and Table 3). The lowest-affinity peptide, D19, did not result in a significant thermal stabilisation of Cdc20.

Table 3.

Figure 6.

D-box peptides inhibit APC/C^Cdc20 ubiquitination activity

We next assessed whether D21 and D20, the two highest affinity peptides, are able to inhibit APC/C^Cdc20 activity. In the context of Cyclin B1 ubiquitination, we found that both peptides are more potent inhibitors compared with Apcin at the same concentration despite having slightly lower Cdc20-binding affinities than Apcin (Fig. 7).

Figure 7.

D-box peptides are able to target mNeon for degradation

To probe the functionality of the D-box variants at the cellular level, we conducted live cell degradation assays using mNeon fusions containing those peptide sequences that contain only natural amino acids: D1, D2, D3, and D19 (Fig. 8A). The D-box sequences were swapped into an RxxL motif previously shown to have no degron activity (Abdelbaki et al. 2022) which we here refer to as ‘D0’, adjacent to the endogenous C-terminal IDR of AURKA to enable processing of the ubiquitinated fusion proteins at the 26S proteasome. We found that all four new D-box variants tested could target mNeon for degradation, with timing consistent with targeting by APC/C^Cdc20 (Fig. 8B). We predicted that the higher affinity D-box peptides from the in vitro assays (D1 and D19) would mediate increased rates and extent of degradation compared to the lower affinity peptides (D2 and D3). However, we found the opposite effect: D2 and D3 showed increased rates of mNeon degradation compared to D1 and D19 (Fig. 8C,D). This observation is consistent with the idea that high-affinity binding at degron binding sites on APC/C, such as in the case of the yeast ‘pseudo-substrate’ inhibitor Acm1, acts to impede polyubiquitination of the bound protein (Qin et al. 2019). Indeed, there is no evidence that Hsl1, which is the highest affinity natural D-box (D1) used in our study, is degraded any more rapidly than other substrates of APC/C in yeast mitosis. As shown in Qin et al., mutation of the high affinity D-box in Acm1 converts it from inhibitor to substrate (Qin et al. 2019). Overall, our results support the conclusions that all the D-box peptides engage productively with the APC/C and that the highest affinity interactors act as inhibitors rather than functional degrons of APC/C.

Figure 8.

Discussion

Here we quantified D-box peptide binding to Cdc20 and show that binding affinities can be enhanced by incorporating unnatural amino acids to better fill the hydrophobic pockets on the Cdc20 surface. We confirmed the success of this approach by determining X-ray crystal structures of Cdc20-peptide complexes. We showed target engagement by the peptides in the cellular context, and we found that the two highest affinity peptides were more potent inhibitors of APC/C^Cdc20 activity than the small molecule Apcin. Lastly, we found that the D-box peptide is a portable motif that can drive productive ubiquitination leading to degradation when fused to a fluorescent protein target.

The finding that the peptides were more potent than Apcin as APC/C^Cdc20 inhibitors was somewhat surprising, since Apcin has a slightly higher Cdc20-binding affinity than the peptides. It suggests that inhibiting APC/C^Cdc20 ubiquitination activity may require larger molecules to compete with substrates effectively. It may also be that, unlike Apcin, the peptides not only block the interaction of substrates with Cdc20 but additionally the interaction with APC10 and/or prevent the conformational change in APC/C that enables recruitment of the E2. In fact, the mechanism of inhibition by Apcin and D-box peptides could be different – it may be that Apcin-bound Cdc20 can still bind to APC/C but peptide-bound Cdc20 cannot. Interestingly, although the inhibitory activity of the D-box peptides roughly correlates with the binding affinity, binding and degradation may be inversely correlated. In addition to the binding of D-box substrates to the co-activators Cdc20 and Cdh1, Qin et al have described how residues C-terminal of the D-box sequence, the ‘D-Box Extension’ DBE motif, influence recruitment of APC10 and potentially APC10 conformational changes enabling the recruitment of the E2 Ube2S (Qin et al. 2019). Ube2S is essential for adding K11 chains, and we showed previously that degradation of all substrates is dramatically slowed down by a lack of Ube2S (Min et al. 2015). The mNeon-D-box constructs used in our current study all contain the same DBE motif, so a potential contribution from this motif will not affect the interpretation of our results, but it could certainly be added as an element in future inhibitor design.

In summary, the finding presented here represent a useful starting point for the further development of APC/C inhibitors as both research tools and also molecular therapeutics. Future directions could involve enhancing potency through avidity by incorporating multiple degrons into our molecules and additions to the D-box core sequence to include motifs that engage other components of the APC/C machinery - namely APC10 and the E2 - thereby not only blocking substrate binding more effectively but also better impeding ubiquitination activity. The results also have implications for the design of small-molecule and peptide-based degraders that harness the APC/C.

Acknowledgements

We acknowledge funding of an AstraZeneca PhD studentship to RSE, and Gates Cambridge Trust and Rosetrees Trust PhD studentships to CO. CL acknowledges funding from BBSRC grant BB/R004137/1 to her lab. We thank Ziguo Zhang for providing the gene encoding Cdc20 in pU1, PIR1 and DH10-multibac^cre cells. We thank David Fischer, Elizabeth Underwood, and Ross Overman for help with insect cell expression and purification of Cdc20, and Jason Breed for help with crystallographic data collection.

Additional files

Supplemental Material