CAND1 inhibits Cullin-2-RING ubiquitin ligases for enhanced substrate specificity

Kankan Wang; Stephanie Diaz; Lihong Li; Jeremy R Lohman; Xing Liu

doi:10.1038/s41594-023-01167-5

. Author manuscript; available in PMC: 2024 Aug 1.

Published in final edited form as: Nat Struct Mol Biol. 2024 Jan 4;31(2):323–335. doi: 10.1038/s41594-023-01167-5

CAND1 inhibits Cullin-2-RING ubiquitin ligases for enhanced substrate specificity

Kankan Wang ¹, Stephanie Diaz ^1,², Lihong Li ^1,³, Jeremy R Lohman ^1,⁴, Xing Liu ^1,^3,^*

PMCID: PMC10923007 NIHMSID: NIHMS1957804 PMID: 38177676

Abstract

Through targeting essential cellular regulators for ubiquitination and serving as a major platform for discovering PROteolysis-TArgeting Chimera (PROTAC) drugs, Cullin-2 (CUL2)-RING ubiquitin ligases (CRL2s) comprise an important family of CRLs. The founding members of CRLs, the CUL1-based CRL1s, are known to be activated by CAND1, which exchanges the variable substrate receptors associated with the common CUL1 core and promotes the dynamic assembly of CRL1s. Unexpectedly, here we find that CAND1 inhibits CRL2-mediated protein degradation in human cells. This effect arises due to altered binding kinetics, involving CAND1 and CRL2^VHL, as we illustrate that CAND1 dramatically increases the dissociation rate of CRL2s but hardly accelerates the assembly of stable CRL2s. Using PROTACs that differently recruit neo-substrates to CRL2^VHL, we demonstrate that the inhibitory effect of CAND1 helps distinguish target proteins with different affinities for CRL2s, presenting a mechanism for selective protein degradation with proper pacing in the changing cellular environment.

Keywords: ubiquitin ligases, VHL, Cullin-2, CAND1, PROTAC, neddylation, COP9 signalosome, selective protein degradation, binding kinetics, dynamics of protein complex

INTRODUCTION

The ubiquitin-proteasome system (UPS) represents a crucial regulatory machinery determining the half-lives of numerous proteins in eukaryotic cells. Through sequential enzymatic reactions, protein targets are covalently linked with ubiquitin or ubiquitin chain, the latter leading to 26S proteasome-mediated degradation¹. In the UPS, multi-subunit cullin-RING ligases (CRLs) constitute the largest family of E3 ubiquitin ligases that recognize protein substrates and couple them with activated ubiquitin deposited by E2-conjugating enzymes^2,3. All CRLs contain a cullin-RING enzymatic core, and humans have six canonical cullin members (CUL1/2/3/4A/4B/5). The six cullins can assemble into hundreds of ubiquitin ligases with similar molecular architectures by binding substrate receptor modules typically consisting of an adaptor and interchangeable substrate receptor^4,5. Specially, the CUL2-RING ligase (CRL2) is composed of the CUL2 scaffold, the RING-domain protein RBX1, the adaptor comprising ELOB/ELOC proteins, and a substrate receptor protein containing a BC-box motif that binds the adaptor^6-8. The human genome encodes >30 BC-box proteins, >ten of which have been shown to form CRL2s^8,9. The CRL2^VHL, which carries VHL as the substrate receptor, is a well-understood CRL2 due to its familiar substrate HIF-1α, a master regulator of hypoxia-induced responses in cells^10,11. In addition, using molecular glue compounds or PROteolysis TArgeting Chimeras (PROTACs) to induce endogenous E3 ligases to target disease-causing proteins for degradation has emerged as a promising strategy for drug discovery. Recently, many PROTAC protein degraders designed on VHL have achieved targeted protein degradation via CRL2^{VHL 12,13}.

CRLs are regulated by master regulators—such as NEDD8, CSN, and CAND1—that either interact with or change post-translational modification of the cullin-RING core. NEDD8, a ubiquitin-like protein, conjugates to the C-terminal domain of cullins (neddylation) and triggers CRL-E2 complex conformational changes, resulting in CRL activation and enhanced CRL-mediated ubiquitination^14-17. The NEDD8-cullin conjugation can be cleaved by CSN, a nine-subunit complex with its metalloprotease catalytic domain carried by the CSN5 subunit^18,19. While NEDD8 removal by CSN (de-neddylation) deactivates CRLs biochemically, fully-active CSN is required for proper CRL function in vivo^20-24, partly because CSN regulates CAND1 activity. CAND1 is a cullin-binding protein whose working mechanism is based on CUL1-RING ligases (CRL1s, alternatively CUL1•SKP1•F-box-protein or SCF ubiquitin ligase). The horseshoe shaped CAND1 clamps around CUL1 and buries the neddylation site. Thus, CAND1-bound CUL1 cannot be neddylated, and CAND1 cannot stably bind neddylated CUL1^25-27. When CAND1 associates with unneddylated CUL1 in a CRL1 complex, a CAND1-SKP1 adaptor clash catalyzes replacement of the substrate receptor modules assembled on the same CUL1 scaffold. This CAND1-CRL1 interaction feature defines CAND1 as an exchange factor, which promotes the exchange of diverse CUL1 substrate receptor modules in human cell lysate^28-31. In living human cells, CAND1-mediated exchange enables dynamic assembly of CRL1s, allowing CUL1 recruitment to substrate receptors loaded with substrates “just-in-time” for target protein ubiquitination and degradation^28,29,32. Because of this exchange mechanism, CAND1 activates CRL1s, and cells lacking CAND1 are defective in degrading CRL1 substrates^28,29,33-39.

Through engaging different groups of substrate receptor modules, other cullins assemble CRLs with varying degrees of structural difference from CRL1s^4,14,40. Among them, CAND1 seems to regulate CRL3/4s similarly to CRL1s. Depleting CAND1 from human cells resulted in impaired degradation of NRF2, a CRL3 substrate⁴¹, and RBM39, a CRL4 neo-substrate in the presence of the molecular glue indisulam⁴². In addition, CAND1 promotes the exchange of multiple substrate receptor modules for CUL4B in human cell lysate⁴². Furthermore, a CRISPR screen for loss-of-function mutations in leukemia cells that confer drug resistance to CRL4-based protein degraders hit genes encoding CAND1, CSN subunits, and CUL4 substrate receptors that the degraders bind to⁴³. Similar results were independently obtained using myeloma cells⁴⁴, supporting the understanding that both CAND1 and CSN activates CRL4-mediated protein degradation. In contrast and interestingly, the CRISPR knockout screen for drug resistance to ARV-771, a CRL2^VHL-based PROTAC protein degrader, hit the gene encoding VHL but neither CAND1 nor CSN subunits⁴³. This result indicates CAND1 differentially regulates CRL2^VHL. Using biochemical, biophysical, and cell biological approaches, we performed quantitative studies on CAND1-CRL2 interactions and CRL2-mediated protein degradation. Our results uncovered the molecular basis and implied the biological relevance for the inhibitory role of CAND1 in regulating CRL2 assembly and activity in human cells.

RESULTS

CAND1 inhibits the degradation of CRL2 substrates

Given that knocking out CAND1 in human cells failed to stabilize ARV-771 induced degradation of CRL2^VHL neo-substrates⁴³, we first wanted to confirm CAND1-CUL2 interaction. We immunoprecipitated (IP’d) CAND1^HA expressed in CAND1/CAND2 double knockout (DKO) HEK293 cells, and CUL2 was co-IP’d with CAND1^HA (Fig. 1a). Only the band of unmodified CUL2 was detected in the presence and absence of neddylation inhibitor MLN4924⁴⁵, suggesting CAND1 preferentially binds unneddylated CUL2. VHL was not co-IP’d with CAND1^HA, confirming that like CRL1-CAND1 interactions, CUL2 does not form a stable complex with both CAND1 and the VHL•ELOB•ELOC (VBC) substrate receptor module (Fig. 1a). The CAND1-CUL2 interaction was further confirmed in in vitro pulldown assays, where CAND1 bound both versions (full-length and “split’n co-express”⁴⁶) of unneddylated ^StrepIICUL2 (Fig. 1b).

Fig. 1 — a, CAND1 binds unneddylated CUL2 in human cell lysate. Whole cell lysate (WCL) from the indicated HEK293 cell lines was immunoprecipitated (IP) with anti-HA resin followed by immunoblotting with indicated antibodies. GAPDH served as a loading control throughout this work. DKO: *CAND1* and *CAND2* double knockout cells; DKO^CAND1-HA: DKO cells expressing transgenic CAND1^HA.

b, CAND1 binds unneddylated CUL2 *in vitro*. StrepII-tag affinity pulldown assay detecting the ^StrepIICUL2•CAND1 complex in the presence and absence of CUL2 neddylation. In each assay, 1 μM CAND1 was incubated with 0.6 μM RBX1•^StrepIICUL2. ^StrepIICUL2^FL: full-length ^StrepIICUL2; ^StrepIICUL2^split: “split’n co-expressed” ^StrepIICUL2; CTD: C-terminal domain; N8: neddylation.

**c,d,** CAND1 inhibits ARV-771 induced degradation of BRD2, a CRL2^VHL neo-substrate. c, Cell lines in a were immunoblotted with anti-CAND1 antibody. n = 3. WT: wild-type HEK293 cells. d, Cells in c were treated with 50 nM ARV-771 for the indicated time and immunoblotted with anti-BRD2 and anti-GAPDH antibodies. Quantitative analyses of BRD2 degradation were shown on the right. n = 3.

e, Treatment with ARV-771 at various concentrations decreases levels of BRD2 and increases levels of apoptosis marker in DKO cells. Cells were treated with ARV-771 at indicated concentrations for 18 h and immunoblotted using anti-BRD2 and anti-cleaved PARP antibodies. Levels of BRD2 relative to the untreated (UT) control were quantified and shown in the bottom graph. n = 3.

f, CAND1 inhibits the degradation of the CODD degron motif from HIF1α, a CRL2^VHL substrate. Cells expressing ^FLAGCODD pretreated by 200 μM desferrioxamine were subjected to cycloheximide (CHX) chase assay. Quantitative analyses of ^FLAGCODD degradation were shown on the bottom. n = 3. S.E.: short exposure; L.E.: long exposure.

g, CAND1 inhibits the degradation of MIC19, a CRL2^KLHDC2 substrate. Cells expressing ^FLAGMIC19 were treated with 60 μg/ml CHX for the indicated time. Quantitative analyses of ^FLAGMIC19 degradation were shown on the right. n = 3.

All error bars represent SEM. “n” indicates number of independent experiments. P values were determined by two-sided t-tests with no adjustments.

Next, we tested how the CAND1-CUL2 interaction affects CRL2^VHL activity in human cells, by analyzing ARV-771 induced degradation of BRD2, a CRL2^VHL neo-substrate⁴⁷. Unexpectedly, BRD2 degradation became both faster and more complete in DKO cells, with a 34% reduction in t_1/2 for BRD2 elimination and a 42% reduction in the BRD2 plateau level compared to the wild-type (WT) cells. This accelerated degradation of BRD2 was due to the loss of CAND1, because reintroducing CAND1 to DKO cells reverted the degradation to the WT level (Fig. 1c,d). We also noticed an increased level of VHL in the DKO cells (Extended Data Fig. 1a), which might contribute to the enhanced BRD2 degradation. To rule out this possibility, we integrated a VHL transgene into WT cells through the Flp recombinase-mediated insertion (WT^VHL), and we titrated the expression of the transgenic VHL to achieve identical levels of VHL in WT^VHL and DKO cells (Extended Data Fig. 1a). The ARV-771 induced degradation of BRD2 in the WT^VHL cells did not differ from that in WT cells (Extended Data Fig. 1b,c), demonstrating the enhanced degradation of BRD2 in DKO cells was primarily due to the lack of CAND1. Because the plateau level of BRD2 in our degradation assay was lower in the DKO cells (Fig. 1d), we wondered if this was due to an insufficient assay duration and if it was ARV-771 concentration dependent. Thus, we examined the elimination of BRD2 after treating cells with 25-200 nM ARV-771 for 18 h. Consistently, compared to WT cells, BRD2 levels were reduced by 50-60% in DKO cells (Fig. 1e). This increased elimination of BRD2 in DKO cells was physiologically significant because cleaved PARP was detected in ARV-771 treated DKO but not WT cells (Fig. 1e), indicating increased apoptosis in the ARV-771 treated DKO cells.

Our results have demonstrated that CAND1 inhibits CRL2^VHL-mediated degradation of neo-substrates, but it is unclear if the same effect applies to CRL2 natural substrates, such as HIF-1α. We previously reported that the degradation of truncated HIF-1α containing only the C-terminal Oxygen-Dependent Degradation (CODD) domain better reflects CRL2^VHL activity in HEK293 cells than the degradation of full-length HIF-1α, because full-length HIF-1α was primarily degraded by VHL-independent mechanisms in the cycloheximide chase assay⁴⁸. Therefore, we analyzed the degradation of CODD, and like BRD2, CODD was degraded faster in DKO cells, with t_1/2 reduced by 57% compared to WT cells (Fig. 1f). Furthermore, we analyzed degradation of MIC19, a glycine-ended protein previously characterized as a CRL2^KLHDC2 substrate^49,50, and the t_1/2 was reduced by 87% when CAND1 was lost (Fig. 1g).

Effects of neddyation and deneddylation on BRD2 degradation

CAND1 activity is associated with the cullin neddylation status. Thus, we inhibited neddylation or de-neddylation in cells and asked how the ARV-771 induced degradation of BRD2 would be affected. Consistent with our previous conclusion that neddylation is required for CRL2^VHL-mediated degradation of CODD⁴⁸, treating cells with MLN4924 stabilized BRD2. Even when MLN4924 was supplied at a concentration not sufficient to eliminate neddylated CUL2, BRD2 degradation was fully blocked (Fig. 2a). In contrast, treating cells with the CSN inhibitor CSN5i-3⁵¹ mildly enhanced BRD2 degradation in WT cells (Fig. 2b). This result was counterintuitive, because reduced CSN activity is usually associated with CRL inactivation in cellular context^52-54. We rationalized that this might be because through keeping CUL2 neddylated, CSN5i-3 prevents CAND1 from binding and inhibiting CRL2s. Thus, we treated DKO cells with CSN5i-3, which then imposed a mild negative impact on BRD2 degradation, leaving 2.3-fold more BRD2 at the end of the degradation assay (Fig. 2c). The opposite effects of CSN5i-3 on BRD2 degradation in WT and DKO cells aligned with the conclusion that CAND1 inhibits BRD2 degradation. We also analyzed the effect of CSN5i-3 on CODD degradation in WT and DKO cells. The results were generally in consensus with BRD2 degradation (Extended Data Fig. 2).

Fig. 2 — a, CUL2 neddylation is required for ARV-771 induced degradation of BRD2. WT cells pre-treated with the indicated concentrations of MLN4924 for 1 h were incubated with 50 nM ARV-771 for the indicated time and were immunoblotted with anti-BRD2 and anti-CUL2 antibodies. Quantified BRD2 levels relative to the BRD2 level at the zero-time point were shown in the right panel. n = 3.

b, Eliminating de-neddylation in WT cells mildly enhances ARV-771 induced degradation of BRD2. Same as in a except that cells were pre-treated with 1 μM CSN5i-3 for 1 h. Control samples were treated with equivalent amount of DMSO. Quantitative analyses of BRD2 degradation were shown on the right. n = 4.

c, Eliminating de-neddylation in DKO cells mildly affects ARV-771 induced degradation of BRD2. Same as in b except that DKO cells were used. n = 3.

All error bars represent SEM. “n” indicates number of independent experiments. P values were determined by two-sided t-tests with no adjustments.

CAND1 does not promote exchanges of CRL2 substrate receptors

The observation that CAND1 slowed down ARV-771 induced degradation of BRD2 implies that CAND1 is not an exchange factor for CUL2 substrate receptors. We thus decided to assay the exchange of CUL2 ligands in human cell lysate, using a Stable Isotope Labeling with Amino acids in Cell culture (SILAC) assay. We proceeded to fuse a peptide tag to the N-terminus of the endogenous CUL2 for efficient precipitation of CUL2 without interfering the cellular stoichiometry, a strategy previously used for studying other CRLs^28,42. Because VBC binds next to the CUL2 N-terminus^55,56, we first assessed effects of peptide tags on the stability of CRL2^VHL. Using bio-layer interferometry (BLI), we compared the K_D of ^GSTVBC in complex with CUL2^StrepII (control), ^StrepIICUL2, and ^3xFLAGCUL2 (Fig. 3a and Extended Data Fig. 3a,b). Fusing a StrepII tag at the CUL2 N-terminus marginally affected the K_D, while fusing a 3xFLAG acidic tag increased the K_D by 5-fold (Fig. 3b). We therefore used CRISPR/Cas9 and integrated the sequence coding the StrepII tag at the 5’ end of the CUL2 gene in DKO cells, to convert all endogenous CUL2 to ^StrepIICUL2 (Fig. 3c and Extended Data Fig. 3c-f). These DKO^StrepII-CUL2 cells were then grown in “light” culture media, while DKO or DKO^CAND1—the CAND1 complementation cells (Fig. 1c)—were grown in “heavy” culture media containing stable isotope labeled Arg and Lys. To maximize the effect of CAND1, we eliminated CUL2 neddylation with MLN4924, and as a control, we included CSN5i-3-treated cells (Fig. 3d). The “light-” and “heavy-”grown cells were mixed, lysed, incubated for one hour at room temperature to allow protein exchange to occur, and proteins co-precipitated with ^StrepIICUL2 were analyzed by mass spectrometry (Fig. 3d and Extended Data Fig. 3g). Because ^StrepIICUL2 only existed in “light”-grown cells, higher heavy/light ratios of ^StrepIICUL2 ligands indicated higher exchange rates. When CAND1 was absent (Fig. 3e, orange), CUL2 and RBX1 had no detectable exchange as expected. CSN subunits had relatively high exchange rates, at levels comparable to those reported for CUL4⁴². CUL2 substrate receptors showed low to modest exchange rates, with VHL being the most highly exchanged species. Importantly, when CAND1 was present (Fig. 3e, green), exchange rates for all CUL2 substrate receptors were unchanged, consistent with the understanding that CAND1 was not a CRL2 exchange factor.

An interesting and unexpected observation came from the CSN5i-3 treatment (Fig. 3f, purple). When CUL2 was predominantly neddylated (Extended Data Fig. 3g), exchange levels for CSN subunits were reduced by ~50%, consistent with the lower k_off of ^NEDD8CUL•CSN than CUL•CSN⁵⁷. Interestingly, exchange levels for components in substrate receptor modules—such as ELOB, ELOC, VHL and ZYG11B—were also reduced, suggesting ^NEDD8CRL2 had a lower k_off than CRL2. To test this possibility, we monitored the dissociation of VBC from ^NEDD8CUL2 or CUL2 in the DKO^StrepII-CUL2 cell lysate. ^StrepIICUL2 was affinity precipitated after incubation with VHL^114-213 chase for varying time (Fig. 3g and Extended Data Fig. 3h), and levels of VHL that remained bound by ^StrepIICUL2 were determined (Fig. 3h). This analysis revealed that with t_1/2 of 105 min vs. 30 min, dissociation of ^NEDD8CRL2^VHL was slower than CRL2^VHL (Fig. 3i), in line with the finding from the SILAC assay.

Binding kinetics of CRL2^VHL and CUL2•CAND1

The finding that CAND1 functioned as an inhibitor instead of an exchange factor for CRL2 immediately led to the question: how does CAND1 regulate CRL2 and what makes it different from the regulation of CRL1? To seek the answers, we set out to characterize the dynamic interactions involving CUL2, CAND1, and VBC, using Förster resonance energy transfer (FRET) based in vitro assays. First, we sought a FRET assay for CUL2-VBC interaction. Two isoforms of ELOC were used in previous structural studies of CRL2^{VHL 55,56,58}. The short isoform (ELOC^17-112) lacks the N-terminal 16 amino acid residues of the long isoform (ELOC^1-112), an intrinsically disordered region conserved across species (Extended Data Fig. 4a). Using BLI, we found that the K_D of CRL2^VHL (or CUL2•VBC) increased 10-fold when ELOC^1-112 was changed to ELOC^17-112 (Fig. 4a,b and Extended Data Fig. 4b), primarily due to a much higher k_off when ELOC^17-112 was incorporated (Extended Data Fig. 4c,d). We then confirmed our HEK293 cells exclusively expressed ELOC^1-112 (Fig. 4c), and therefore, we co-expressed ELOC^1-112 with ELOB and VHL to generate VBC for all experiments. We labeled VBC with the FlAsH fluorophore (VBC^FlAsH) via a tetracysteine tag at the C-terminus of ELOB, and we fused the cyan fluorescent protein (CFP) to the N-terminus of CUL2 (^CFPCUL2). Mixing ^CFPCUL2 and VBC^FlAsH yielded FRET with 32% donor signal reduction, which was competed away by excess unlabeled VBC (Fig. 4d), confirming the FRET signal depended on ^CFPCUL2-VBC^FlAsH interaction. Using this FRET assay, we monitored the dissociation of ^CFPCUL2•VBC^FlAsH in the presence of excess VBC chase, which gave k_off of 5.7×10⁻⁴ s⁻¹ or t_1/2 of 20 min (Fig. 4e), similar to the k_off determined using human cell lysate (Fig. 3i). To measure k_on, we mixed ^CFPCUL2 with increasing concentrations of VBC^FlAsH in a stopped-flow apparatus and monitored CFP fluorescence over time (Extended Data Fig. 4e). This analysis revealed that the association of CRL2^VHL was biphasic (Fig. 4f), with the first (fast) phase giving k_on,fast of 1.1×10⁶ M⁻¹ s⁻¹ (Fig. 4g). The k_obs in the second (slow) phase appeared independent of ligand concentrations, giving k_on,slow of 0.015 s⁻¹ (Fig. 4h). These features of a biphasic association suggested that the assembly of CRL2^VHL was a two-step event, where conformational changes occurred following the initial binding step to yield the final stable complex.

**a,b,** VHL•ELOB•ELOC^1-112 binds CUL2 more tightly than VHL•ELOB•ELOC^17–112. a, Similar to Fig. 3a,b but ^GSTVBC with different isoforms of ELOC was immobilized. Black lines represent one-phase exponential curve fitting. Relative K_D were shown in b. n = 3.

c, HEK293 cells only express the ELOC^1-112 isoform. HEK293 total cell lysate, recombinant ELOC^17-112 and ELOC^1-112 were immunoblotted with the anti-ELOC antibody.

d, FRET assay for the CUL2•VBC complex. Fluorescence emission spectra from excitation at 430 nm of indicated samples. Chase: 1 μM (10×) VBC. Proteins were added in the indicated order and assayed within 1 min after mixing.

e, Dissociation of the CUL2•VBC complex. Relative CFP fluorescence versus time after addition of VBC chase to preincubated ^CFPCUL2 and VBC^FlAsH. Single exponential fit yielded the k_off. n = 3.

**f-h,** Association of the CUL2•VBC complex. f, Representative two-phase exponential curve fitting (orange) for real-time CFP fluorescence (grey) upon addition of 120 nM VBC^FlAsH to 10 nM ^CFPCUL2 in a stopped-flow fluorimeter. **g,h,** The fast-phase k_obs **(g)** and slow-phase k_obs **(h)** of ^CFPCUL2•VBC^FlAsH association at different concentrations of VBC^FlAsH (Extended Data Fig. 4e) were plotted. Linear slope gave the k_on,fast **(g)**. Y-intercept gave the k_on,slow **(h)**. n = 5.

i, FRET assay for the CUL2•CAND1 complex. Fluorescence emission spectra from excitation at 350 nm of indicated samples. ^FlAsHCAND1 was produced from insect cells. Chase: 1 μM (10×) CAND1. Proteins were added in the indicated order and assayed within 1 min after mixing.

j, Dissociation of the CUL2•CAND1 complex. Relative AMCA fluorescence versus time after addition of CAND1 chase to preincubated CUL2^AMCA and ^FlAsHCAND1. ^FlAsHCAND1 was produced from insect cells. Single exponential fit yielded the k_off. n = 3.

k, Association of the CUL2•CAND1 complex. The k_obs of CUL2^AMCA•^FlAsHCAND1 assembly at different concentrations of ^FlAsHCAND1 (Extended Data Fig. 4j) were plotted. Linear slope gave the k_on. n = 5.

All error bars represent SEM. “n” indicates number of independent experiments. P values were determined by two-sided t-tests with no adjustments.

To establish a FRET assay for CUL2-CAND1 interaction, we first estimated the dissociation rate for CUL2•CAND1 through co-precipitating ^StrepIICUL2 and CAND1^HA from human cell lysate supplemented with ^GSTCAND1 chase, which gave a dissociation t_1/2 of 13 min (Extended Data Fig. 4f,g). Unexpectedly, a similar experiment using recombinant ^GSTCAND1 and CUL2 (with CUL2^split chase) generated from bacterial cells revealed t_1/2 of 3.5 min (Extended Data Fig. 4h,i). The differing CUL2•CAND1 t_1/2 in the two assays may be explained by the lack of post-translational modifications in proteins prepared from bacterial cells, and therefore, we repeated the assay using ^GSTCAND1 made from insect cells (Extended Data Fig. 4h, bottom panel). The t_1/2 of CUL2•CAND1 was then estimated to be 16 min (Extended Data Fig. 4i), close to the t_1/2 measured from human cell lysate samples (Extended Data Fig. 4g). Thus, we purified CAND1 from insect cells and established a FRET assay using CUL2^AMCA and ^FlAsHCAND1 (Fig. 4i), which revealed k_off of 4.7×10⁻⁴ s⁻¹ (Fig. 4j) and k_on of 1.4×10⁸ M⁻¹ s⁻¹ (Fig. 4k and Extended Data Fig. 4j) for CUL2-CAND1 interaction.

CAND1 strongly affects the k_off of CRL2^VHL

We next asked if and how CAND1 affects CRL2^VHL stability. First, we assessed the effect of CAND1 on the association of ^CFPCUL2•VBC^FlAsH by adding increasing concentrations of VBC^FlAsH to ^CFPCUL2 that was pre-assembled with CAND1 (Extended Data Fig. 5a). Unlike the biphasic association of ^CFPCUL2•VBC^FlAsH (Fig. 4f), under this condition, the donor signal fit a one-phase exponential model, giving k_on of 4.8×10⁵ M⁻¹ s⁻¹ (Fig. 5a,b). We then assessed if CAND1 affected the dissociation of ^CFPCUL2•VBC^FlAsH. Adding CAND1 to pre-assembled ^CFPCUL2•VBC^FlAsH immediately abolished the FRET (Fig. 5c,d), suggesting CUL2•VBC was quickly disassembled when CAND1 was present. The similar effect of CAND1 on CUL2•VBC disassembly was also observed when ^StrepIICUL2•VBC was affinity precipitated from human cell lysate with or without CAND1 (Extended Data Fig. 5b,c). In addition, adding CAND1 with two amino acid substitutions in the β-loop (CAND1β++), a mutant CAND1 incapable of disassembling CRL1³¹, did not result in ^CFPCUL2•VBC^FlAsH disassembly (Extended Data Fig. 5d). The CAND1 β-loop was shown to clash with the SKP1 adaptor in CRL1^27,31,59, and similarly, a clash between the β-loop and ELOC was observed when CUL1•CAND1 and CRL2^VHL structures were superimposed (Extended Data Fig. 5e). Thus, it appears that the same mechanism applies to CAND1-mediated disassembly of both CRL1 and CRL2. Next, we quantified the rate of CAND1-mediated dissociation of ^CFPCUL2•VBC^FlAsH, and we monitored the real-time CFP signal change triggered by the addition of CAND1. Higher concentrations of CAND1 resulted in faster dissociation (k_obs) of ^CFPCUL2•VBC^FlAsH (Extended Data Fig. 5f), with a maximum rate at 0.15 s⁻¹ and a half maximal concentration (K_M) of 60 nM (Fig. 5e). This result is comparable to the previous finding of CAND1-mediated disassembly of CRL1³⁰. Like the CRL1-CAND1 interaction, the saturation kinetics here indicate the presence of a VBC•CUL2•CAND1 ternary complex as a rate-limiting state, and the maximal observed rate of 0.15 s⁻¹ represents the k_off of VBC^FlAsH from the ternary complex. Thus, CAND1 accelerates CRL2^VHL disassembly, increasing the k_off from 5.7×10⁻⁴ s⁻¹ to 0.15 s⁻¹.

**a,b,** Association of CUL2•VBC in the presence of CAND1. a, Representative result showing the one-phase association of ^CFPCUL2•VBC^FlAsH in the presence of CAND1. Real-time detection of CFP fluorescence (grey) in a stopped-flow fluorimeter upon addition of 120 nM VBC^FlAsH to 10 nM ^CFPCUL2 preassembled with 20 nM CAND1. Signal changes were fit to a one-phase exponential curve (orange). b, k_on for VBC binding to CUL2 preassembled with CAND1. The k_obs of ^CFPCUL2•VBC^FlAsH association at different concentrations of VBC^FlAsH (Extended Data Fig. 5a) were plotted. Linear slope gave the k_on. n = 5 for all concentrations except for 40 nM (n = 4). CAND1 for the entire Fig. 5 was produced from insect cells.

**c-e,** CAND1 significantly increases the k_off of the CUL2•VBC complex. c, As in Fig. 4d except with 100 nM CAND1 added in the presence of VBC chase. The protein mixtures were incubated for 1 min before analyses. d, Fast disassembly of CUL2•VBC by CAND1. Real-time detection of CFP fluorescence in a stopped-flow apparatus upon addition of 20 nM CAND1 to 10 nM ^CFPCUL2 preincubated with 10 nM VBC^FlAsH. Chase: 100 nM VBC. e, Single exponential observed dissociation rates (k_obs) for increasing concentrations of CAND1 mixed with 10 nM ^CFPCUL2 preincubated with 10 nM VBC^FlAsH (Extended Data Fig. 5f). Hyperbolic curve fitting gave the k_off and K_M. n = 5.

All error bars represent SEM. “n” indicates number of independent experiments.

VBC alone marginally affects the k_on or k_off of CUL2•CAND1

We subsequently examined effects of VBC on CUL2•CAND1 assembly and disassembly. ^FlAsHCAND1 associated with CUL2^AMCA preassembled with VBC with k_on of 9.8×10⁷ M⁻¹ s⁻¹ (Fig. 6a and Extended Data Fig. 6a-c), which was almost identical to the k_on in the absence of VBC (Fig. 4k). What captured our attention was that upon mixing VBC with CUL2^AMCA•^FlAsHCAND1 in the presence of CAND1 chase, FRET was only mildly reduced (Fig. 6b), indicating VBC has no major impact on CUL2•CAND1 disassembly. Consistent with this finding, monitoring AMCA signal upon mixing VBC with CUL2^AMCA•^FlAsHCAND1 and CAND1 chase revealed almost constant AMCA intensity over the duration of the experiment (Fig. 6c). Besides the FRET-based assays, we also performed a pulldown assay for ^StrepIICUL2•CAND1 dissociation in the presence of ^GSTCAND1 chase (Fig. 6d). Again, VBC showed little effect on the t_1/2 of ^StrepIICUL2•CAND1 (Fig. 6e), confirming VBC minimally promotes the dissociation of CUL2•CAND1.

a, k_on for CAND1 binding to CUL2 preassembled with VBC. Similar to Fig. 4k except that 5 nM CUL2^AMCA preincubated with 20 nM VBC was mixed with increasing concentrations of ^FlAsHCAND1 (Extended Data Fig. 6a). Linear slope gave the k_on. n = 5. CAND1 in the entire Fig. 6 was produced from insect cells.

**b-e,** VBC hardly accelerates the disassembly of CUL2•CAND1. b, As in Fig. 4i except with 100 nM VBC added in the presence of CAND1 chase. The protein mixtures were incubated for 1 min before analyses. c, Disassembly of CUL2^AMCA•^FlAsHCAND1 was unaffected by VBC. Real-time AMCA fluorescence upon addition of 10 nM VBC to 5 nM CUL2^AMCA preincubated with 5 nM ^FlAsHCAND1 in a stopped-flow fluorimeter. Chase: 100 nM CAND1. **d,e,** CUL2•CAND1 dissociation in the presence of VBC **(d)** was analyzed using immunoblotting (e, left panel), and normalized CAND1/^StrepIICUL2 intensity ratios were fit to a single exponential curve to obtain t_1/2 (e, right panel). n = 3.

**f,g,** Neddylation increases the k_off of CUL2•CAND1 in the presence of VBC. f, Relative AMCA fluorescence versus time after addition of the neddylation reaction mix, 200 nM VBC, and 2 μM CAND1 chase to 100 nM CUL2^AMCA preincubated with 100 nM ^FlAsHCAND1. Single exponential fit yielded the k_off. n = 3. g, ^StrepIICUL2•CAND1 was preassembled and mixed with indicated components. Proteins co-precipitated with ^StrepIICUL2 were immunoblotted with indicated antibodies.

h, Immobilized ^StrepIICUL2 or ^StrepIICUL2•VBC was incubated with CAND1. Samples were immunoblotted with indicated antibodies.

i, Same as in h except that immobilized ^StrepIICUL2 or ^StrepIICUL2•CAND1 complex were incubated with VBC.

j, Same as in Fig. 5c except that VBC^FlAsH was replaced by ZYG11B•ELOB^FlAsH•ELOC (ZBC^FlAsH).

k, Same as in b except that VBC was replaced by ZBC.

l, Regulation of CRL1 vs. CRL2 by CAND1.

All error bars represent SEM. “n” indicates number of independent experiments.

In cells, CUL•CAND1 assembly and disassembly are regulated by neddylation. In the CAND1-CRL1 interaction, CUL1 is immediately neddylated when CAND1 is removed and a new CRL1 is assembled, a process driving the cycling of CUL1²⁹. We then wonder if neddylation also plays a role in the CAND1-CRL2 interaction. Thus, we assayed CUL2^AMCA•^FlAsHCAND1 dissociation in the presence of VBC and cullin neddylation. A control group with no NEDD8 E1 enzyme was also included. Consistent with the other measurements, in the presence of VBC but no neddylation, k_off of CUL2^AMCA•^FlAsHCAND1 was unaffected (Fig. 6f, blue group vs. Fig. 4j). However, when neddylation was introduced, the k_off was increased by 5-fold, and the t_1/2 was decreased from 20 min to 4 min (Fig. 6f). These values are echoed by the rate of CUL2 neddylation in HEK293 cells after de-neddylation was blocked by CSN5i-3 (Extended Data Fig. 6d). Additionally, results from pulldown assays also confirmed that the presence of both VBC and neddylation promoted ^StrepIICUL2•CAND1 disassembly (Fig. 6g).

CAND1 promotes CRL2 disassembly but not CRL2 assembly

The kinetic parameters we obtained for CAND1-CRL2 interactions revealed a major difference in the regulation of CRL1 vs. CRL2 by CAND1. While CAND1 increases the k_off of both CRL1 and CRL2, the substrate receptor module increases the k_off of CUL1•CAND1 but not CUL2•CAND1. So similar to CRL1, CAND1 can instantly disassemble a CRL2 and form a stable CUL2•CAND1 complex. However, the immediate formation of a new CRL2 using CUL2 recycled from CUL2•CAND1 was not allowed, which could explain why CAND1 was not an exchange factor for CRL2s (Fig. 3). This idea was supported by results from two in vitro pulldown assays. When CAND1 was added to preassembled and immobilized ^StrepIICUL2•VBC, CAND1 bound ^StrepIICUL2 efficiently while VBC fell off and entered the flowthrough (Fig. 6h). In turn and with a similar experimental setup, when VBC was added to preassembled and immobilized ^StrepIICUL2•CAND1, no CAND1 was detected in the flowthrough and the assembly of ^StrepIICUL2•VBC was inefficient (Fig. 6i). We further changed VBC to ZYG11B•ELOB•ELOC (ZBC) in the FRET and pulldown assays, and we obtained the same results (Fig. 6j,k and Extended Data Fig. 7). Taken together, we conclude that CAND1 inhibits CRL2 because it facilitates only the disassembly but not the assembly of CRL2 complexes (Fig. 6l).

CAND1 enhances CRL2 substrate specificity

The understanding of how CAND1 regulates CRL2 at the biochemical level drove us to the next question: how does the CAND1-mediated disassembly of CRL2 contribute to the selective protein degradation in cells? We rationalized that with accelerated CRL2 disassembly, substrates have to bind CRL2 with high enough affinity to become ubiquitinated before the CRL2 is disassembled. Thus, we hypothesized that the inhibitory effect of CAND1 on CRL2 had a greater impact on the degradation of substrates that bind CRL2 with lower affinities. To test this hypothesis, we took advantage of VHL-based PROTACs that recruit neo-substrates to CRL2^VHL with different affinities (Extended Data Fig. 8a). We first examined the MZ-1 induced degradation of BRD2, BRD3, and BRD4. In the presence of MZ-1, VHL forms the most stable complex with BRD4 and the least stable complex with BRD3^60,61. When WT and DKO cells were treated with 1 μM MZ-1, all three BRD proteins were degraded faster in DKO cells (Fig. 7a). Interestingly, t_1/2 for BRD3 in the WT and DKO cells differed by 3.8-fold, whereas the t_1/2 differed by 1.7-fold for BRD2 and 1.4-fold for BRD4 (Fig. 7a). Thus, CAND1 showed the greatest inhibitory effect on the degradation of BRD3, the substrate with the lowest affinity for CRL2^VHL. Motivated by this result, we further altered the substrate-VHL binding affinity by changing the PROTAC to SIM1. Same as MZ-1 but inducing stronger interactions, SIM1 recruits BRD proteins to CRL2^VHL for degradation, with BRD3 being the least preferred substrate⁶². When WT and DKO cells were treated with 1 μM SIM1, t_1/2 for BRD3 differed by 1.8-fold in the two cell lines (Fig. 7b), a smaller difference than that induced by MZ-1 (Fig. 7a). Moreover, t_1/2 for BRD2 or BRD4 showed almost no difference between the WT and DKO cells (Fig. 7b), a result still consistent with our hypothesis. We further rationalized that through inhibiting the degradation of proteins with low affinities for CRL2, CAND1 might benefit selective protein degradation by increasing the specificity of the CRL2 system. This idea was supported by results showing that BRD3 was only degraded in the absence of CAND1 when cells were treated with a lower concentration of MZ-1 (Extended Data Fig. 8b).

Fig. 7 — a, MZ-1 induced degradation of bromodomain proteins by CRL2^VHL in human cells with or without CAND1. WT or DKO cells were treated with 1 μM MZ-1 for the indicated time and immunoblotted with indicated antibodies (left panel). Quantitative analyses of BRD2, BRD3, and BRD4 degradation were shown in the right panel. n = 3.

b, As in a except that 1 μM SIM1 was used in place of 1 μM MZ-1. n = 3.

c, Degradation of ^3xFLAGGFP fused with a C-degron by CRL2^KLHDC3 in human cells with or without CAND1. WT and DKO cells expressing ^3xFLAGGFP fused with the WT C-degron (WRLTGFSGMKG) were treated with 60 μg/mL CHX for the indicated time and immunoblotted with anti-FLAG antibody. The asterisk indicates GAPDH bands. Quantitative analyses of ^3xFLAGGFP degradation were shown in the bottom panel. n = 3.

d, As in c except that a K/A mutation was introduced to the C-degron for reduced binding affinity with KLHDC3. Quantitative analyses of ^3xFLAGGFP degradation were shown in the bottom panel. n = 3. #: statistical significance. P = 0.004, 0.014, 0.008, 0.007 and 0.003 from left to right.

All error bars represent SEM. “n” indicates number of independent experiments. P values were determined by two-sided t-tests with no adjustments.

Finally, we wondered if our finding on the role of CAND1 in regulating CRL2^VHL-mediated degradation applies to other CRL2 E3 ligases. Thanks to previous works characterizing the C-terminal degrons^49,63-65, we were able to alter substrate affinities and examine the substrate degradation mediated by CRL2^KLHDC3. By fusing the C-degron from SEPHS2 to the C-terminus of ^3×FLAGGFP, we generated a “WT” substrate for CRL2^KLHDC3. We then introduced a K-to-A mutation in the C-degron that was known to impair CRL2^KLHDC3-mediated degradation⁵⁰ and asked how the degradation of this mutant substrate would respond to the loss of CAND1. In the cycloheximide chase assay, the WT substrate was degraded at similar rates in WT and DKO cells (Fig. 7c), which resembled the SIM1-induced BRD2 degradation (Fig. 7b). In contrast, the mutant substrate was only degraded in the DKO cells (Fig. 7d), which resembled the BRD3 degradation at a lower MZ-1 concentration (Extended Data Fig. 8b). Collectively, these results point out that CAND1 helps distinguish substrates with different binding affinities and increase substrate specificity for CRL2s including CRL2^VHL and CRL2^KLHDC3.

DISCUSSION

We have established that CAND1 is a conditional inhibitor for CRL2-mediated protein degradation dependent on CRL2 substrate affinity. In contrast, CAND1 has been known as a CRL1 activator by exchanging substrate receptor modules and allowing quick assembly of new CRL1s. We can explain the differential CAND1 impact on CRL subfamilies, by quantifying binding kinetics for interactions involving CAND1, CUL2, and the VBC substrate receptor module. While the CAND1-CRL1 and CAND1-CRL2 interactions share multiple similarities, there is an essential difference: CAND1 disassembles both CRL1 and CRL2—a process well depicted by two recent structural studies^31,59—but only the CRL1 substrate receptor module can disassemble the CUL•CAND1 complex to instantly reuse the CUL core. Therefore, CAND1 destabilizes existing CRL2s without benefiting new CRL2 assembly, leading to overall inhibition of CRL2 assembly. In addition to in vitro biochemical analyses with CRL2^VHL and CRL2^ZYG11B, our SILAC assay using human cell lysates further demonstrated the lack of accelerated exchange by CAND1 for VHL, ZYG11B, and all other detected substrate receptors.

Our binding kinetics data also revealed CRL2^VHL is much shorter-lived than CRL1. Once assembled, spontaneous dissociation of CRL1^FBXW7 is rare (t_1/2 = 9 days), whereas t_1/2 for CRL2^VHL is 20 min. The CRL2^VHL k_off measured in vitro (5.7 × 10⁻⁴ s⁻¹, Fig. 4e) is similar to measurement in DKO cell lysate (3.8 × 10⁻⁴ s⁻¹, Fig. 3i), implying CRL2s are not regulated by other potential cellular exchange factors. Faster dissociation of VBC from CUL2 may explain why an exchange factor is unnecessary for the CRL2 system. Unexpectedly, we found CUL2 neddylation increased the t_1/2 of CRL2^VHL by 3.5-fold (Fig. 3i) and increased the stability of various CRL2s (Fig. 3f). In light of the current structural model of a ^NEDD8CRL1•E2 complex¹⁵, it is possible that similar to CRL1s, neddylation induces additional protein-protein interactions within ^NEDD8CRL2•E2, where E2 binds both NEDD8 and the substrate receptor, thereby increasing the CRL2 stability.

A feature defining CAND1 as a CRL2 inhibitor is the inability of VBC to disassemble CUL2•CAND1. Then what VBC-CUL2 interaction characteristic differs it from CRL1? Monitoring the real-time association of CRL2^VHL with/without CAND1 provided hints. Association of CRL2^VHL was biphasic in the absence of CAND1 (Fig. 4f-h), indicating formation of a less stable complex followed by stabilization. When CUL2 was bound by CAND1, interactions between VBC and CUL2 still occurred (Fig. 5a), but with a monophasic association displaying k_on ~50% lower than the k_on,fast of CRL2^VHL without CAND1. This association profile suggests that when CUL2 is bound by CAND1, VBC only forms the unstable complex with CUL2 and is unable to remove CAND1 from CUL2. As such, if an additional factor exists to either strengthen the VBC-CUL2 interaction or weaken the CUL2-CAND1 interaction, CAND1 can then be removed, forming a stable CRL2^VHL. Consistent with this expectation, we found adding neddylation to the VBC-CUL2•CAND1 interaction increased the k_off of CUL2•CAND1. Because CUL2 neddylation was enabled immediately after CAND1 dissociation, this k_off (2.8×10⁻³ s⁻¹, Fig. 6f) could be equivalent to the CUL2 neddylation rate. Indeed, measurement of CUL2 neddylation in human cells yielded a similar k_obs (3.4×10⁻³ s⁻¹, t_1/2 = 3.4 min, Extended Data Fig. 6d). This CUL2 neddylation k_obs revealed that in human cells, CUL2•CAND1 was more stable than CUL1•CAND1 (t_1/2 = 0.9 min)²⁹.

Our findings explain previous CRISPR screen results, where CAND1 and CSN were identified as positive regulators of CRL4 but not CRL2^VHL-based protein degraders⁴³. In the CRL4 system, CSN de-neddylates CUL4 to protect the substrate receptors from auto-ubiquitination^21,66 and allow CAND1 to exchange substrate receptors⁴², each of which increases CRL4 activity. For CRL2^VHL, however, CUL2 de-neddylation enables CAND1-mediated CRL2 disassembly, and therefore, CSN may impose a net negative impact on CRL2 activity. This interpretation was supported by results that in WT cells, inhibiting CSN mildly promoted ARV-771 induced BRD2 degradation, while in DKO cells, inhibiting CSN had a mild negative impact (Fig. 2b,c). Following this rationale, if for a certain CRL2 the positive impact from CSN outweighed the negative impact from CAND1, then inhibiting CSN would reduce CRL2-mediated protein degradation in both WT and DKO cells.

Lastly, we explored the role of CAND1 in regulating CRL2 in cells. We proposed and found evidence for the hypothesis that CAND1-mediated disassembly of CRL2 helps distinguish substrates with different affinities. Through repressing de-neddylation, substrate binding stabilizes the CRL complex against the effect of CAND1^28,29. When the bound substrate has a higher k_off, CRL2 has a greater chance to be de-neddylated and disassembled by CAND1, resulting in inefficient substrate ubiquitination. This effect may enhance substrate specificity for the CRL2 system. When selected by CRLs for ubiquitination, protein targets often require post-translational modifications or attachment of molecular glues, which are essentially mechanisms increasing affinities of the protein targets for corresponding CRLs. Furthermore, recent characterization of C-degrons revealed a spectrum of sequences conferring diverse degradation efficiencies by CRL2s⁵⁰. Thus, the accelerated CRL2 disassembly by CAND1 may serve as a mechanism to achieve protein degradation with high selectivity and proper pacing in the changing cellular environment.

To test the “substrate affinity distinguishing” hypothesis, we primarily used PROTAC-induced protein degradation. The neo-substrates in these experiments bind VHL with t_1/2 ranging from seconds to minutes. Therefore, they may quickly move between free vs. CUL2-bound VHL, which could be another basis for their efficient degradation in DKO cells. What if the substrate and receptor forms a very stable complex, like phospho-IκBα•β-TrCP (t_1/2 = 5.9 h)²⁹? In this scenario, the substrate bound to free substrate receptor has to recruit CUL2 to be ubiquitinated. Because t_1/2=20 min for CRL2^VHL and t_1/2=3-4 min for CUL2•CAND1 in the presence of VBC and neddylation, it may be faster to assemble a new CRL2 using the CUL2 released from CUL2•CAND1 than from another CRL2. Based on this rationale, CAND1 may have a positive effect on the degradation of a substrate with very high affinity for CRL2. When such substrates are identified, it may be worth investigating the effect of CAND1 on their degradation rates.

METHODS

Cell lines

For DNA plasmid amplification, DH5α and SURE2 E. coli strains were used. For baculovirus production and amplification, Sf9 insect cells (Thermo Fisher Scientific; Cat. No: 11496015) were used. For protein expression, BL21(DE3) E. coli cells and Sf21 insect cells (Thermo Fisher Scientific; Cat. No: 11497013) were used. For human cell-based assays, Flp-In T-Rex 293 cells (Thermo Fisher Scientific; Cat. No: R78007; RRID: CVCL_U427) were used, and the CAND1/2 DKO cells were generated and confirmed previously²⁹.

Generation of stable cell lines

To generate DKO^StrepII-CUL2 cells, the DNA fragment encoding the StrepII tag was inserted at the 5’ end of the endogenous CUL2 genomic sequence through CRIPSR/Cas9-mediated homology directed repair (HDR). The following gene editing components were electroporated into DKO cells using the Gene Pulser Xcell System (Bio-Rad): 30 μg px330 plasmid encoding Cas9 and gRNA targeting the CUL2 5’ end (gRNA sequence: GAGTAGATTTTGATGAAACA); 60 μg single-stranded oligodeoxynucleotides (ssODN) HDR template containing the coding sequence of the StrepII tag and a linker flanked by a 47 nt 5’ homology arm and a 78 nt 3’ homology arm (5’-GTTATTGTTTTTTAATTGACAGATTTCAACACTACACTTGCACAATGTGGAGCCACCCGCAGTTCGAAAAAGGCGGAGGCTCCGGAGGCGGGAGCGGTACCTCACTTAAGCCCCGTGTTGTTGACTTCGACGAGACATGGAACAAACTTTTGACGACAATAAAAGCCGTGGTCATGTTG-3’) (synthesized by IDT). Twenty-four hours after co-transfection, cells were transferred to 15-cm plates with a density of 300 cells per plate. After approximately 12 days, colonies were isolated and screened for insertion of the StrepII tag by immunoblotting with anti-CUL2 and anti-StrepII antibodies. Colonies showed ^StrepIICUL2 signal were further confirmed by PCR with two primer sets. The first primer set (5’-CAGATTTCAACACTACACTTGCAC-3’ and 5’-AAAGAGCAGCACCCACACAGGTGC-3’) was used to specifically detect the knock-in product. The second primer set (5’-GGAGCTTGCAGTGAGCCAAGATC-3’ and 5’-ATAGTTACATGGGAAAGTTCTTCC-3’) was used to determine the heterogeneity of the edited gene. Lastly, PCR products covering the target site, amplified from a third primer set (5’- GTAAATAGTCTGTGAAATTTCTTGTC -3’ and 5’-AAAGAGCAGCACCCACACAGGTGC-3’), were sequenced to confirm that the incorporated DNA fragment was inserted correctly at both junctions.

The coding sequence of VHL, CAND1^HA, ^3xFLAGODD or ^3xFLAGGFP^CTT was integrated into the Flp-In T-REx 293 cells according to instructions from the manufacturer (Thermo Fisher Scientific). Briefly, pcDNA5/FRT/TO vector containing the gene insert and the pOG44 vector (Invitrogen) were co-transfected into cells using Lipofectamine 3000 (Invitrogen). Cells were then selected using 100 mg/mL hygromycin and confirmed for expression of integrated genes in the presence of tetracycline.

Materials and plasmids

The small molecule compounds, including MLN4924, CSN5i-3, ARV-771, MZ-1 and SIM1 (MedChemExpress), were dissolved in dimethyl sulfoxide (DMSO) to generate stock solutions. Desferrioxamine (Thermo Fisher Scientific) and cycloheximide (DOT Scientific) were dissolved in water to generate stock solutions. Tetracycline was dissolved in 70% ethanol to generate a stock solution.

The following constructs were generated for expressing recombinant protein in bacterial cells. The coding sequences of ELOB and ELOC were codon optimized, synthesized (Gene Universal), and integrated into the pACYCDuet-1 vector (Novagen) via restriction digestion and ligation. The DNA fragment encoding codon optimized ^17-112VHL was synthesized (Gene Universal) and inserted into the pGEX-4T vector (Cytiva) via restriction digestion and ligation. The DNA fragment encoding ZYG11B was synthesized and inserted into the pGEX-4T vector (Genscript). Using the Site-Directed Mutagenesis Kit (New England Biolabs), a Lumio tag (TGTTGTCCAGGATGTTGT) was attached to the 3’ end of the ELOB gene, and a sortase tag (TTACCAGAAACAGGAGGA) was attached to the 3’ end of the CUL2 gene. The construct encoding the ELOC long isoform (ELOC^1-112) was generated by adding the coding sequence for the N-terminal 16 amino acids via site-directed mutagenesis. The construct encoding the truncated VHL (VHL^114-213) was similarly generated by the Site-Directed Mutagenesis Kit. Constructs expressing recombinant CUL2•RBX1 were generated and confirmed previously⁴⁶. The construct for expressing full-length CUL2•RBX1 was used to express ^CFPCUL2•RBX1, where the coding sequence of CFP was fused to the 5’ end of the CUL2 gene using the HiFi DNA Assembly Cloning Kit (New England Biolabs). The construct for expressing CAND1β++ mutant was derived from the wild-type version³⁰ using the Site-Directed Mutagenesis Kit.

For expressing recombinant protein using the Baculovirus system, the coding sequence of GST-thrombin-CAND1 or GST-thrombin-^Lumio-tagCAND1 was integrated into the pVL1393 vector (Invitrogen). For CRIPSR/Cas9-mediated knock-in in HEK293 cells, the gRNA was inserted into px330 backbone as described previously. For generating stable HEK293 cell lines using the Flp-In system, the coding sequence of VHL, CAND1^HA or ^3xFLAGODD was inserted into the pcDNA5/FRT/TO vector (Thermo Fisher Scientific) via restriction digestion and ligation. The CHCHD3 cDNA (coding the MIC19 protein) clone (a kind gift from Dr. Stephen J. Elledge) was assembled into a pcDNA5/FRT/TO vector and fused with a N-terminal FLAG-HA tag via the HiFi DNA Assembly Cloning Kit (New England Biolabs).

Antibodies

The following primary antibodies (1:1000 dilution, unless specified) were used in Western blot: anti-CAND1 (Bethyl Laboratories # A302-901A; Santa Cruz Biotechnology sc-10672, 1:2000 dilution), anti-CAND2 (Bethyl Laboratories # A304-046A), anti-GAPDH (Santa Cruz Biotechnology sc-47724, 1:5000 dilution; Cell Signaling # 2118), anti-CUL2 (Thermo Fisher Scientific # 51-1800; Santa Cruz Biotechnology sc-166506), anti-VHL (Cell Signaling # 68547), anti-ELOC (Santa Cruz Biotechnology sc-135895, 1:500 dilution), anti-ELOB (Bethyl Laboratories # A304-008A), anti-BRD2 (Cell Signaling # 5848), anti-BRD3 (Abcam # ab50818, 1:500 dilution), anti-BRD4 (Abcam # ab128874), anti-FLAG (Sigma-Aldrich # F1804, 1:5000 dilution), anti-StrepII (Abcam # ab76949, 1:500 dilution), anti-HA (Cell Signaling # 3724), anti-cleaved PARP (Asp214) (Cell Signaling # 5625), HRP-conjugated anti-FLAG (Sigma-Aldrich # A8592). Secondary antibodies (1:10000 dilution) used for the Western blot include Alexa Fluor 790 conjugated anti-mouse IgG (Abcam # ab186699), Alexa Fluor 790 conjugated anti-rabbit IgG (Thermo Fisher Scientific # A11374), Alexa Fluor 680 conjugated anti-rabbit IgG (Abcam # ab175772), and Alexa Fluor 680 conjugated anti-mouse IgG (Thermo Fisher Scientific # A10038).

Protein expression and purification

The expression and purification of CUL2•RBX1 and ^CFPCUL2•RBX1 followed the protocol described previously^46,67. Briefly, the protein complex was expressed in BL21 (DE3) cells, purified by Ni-affinity chromatography followed by digestion with the TEV protease, and further purified by StrepII-tag affinity chromatography and size exclusion chromatography. To generate CUL2^AMCA•RBX1, CUL2^sortase-tag•RBX1 was expressed in BL21 (DE3) cells, purified by Ni-affinity chromatography, TEV protease digestion, and StrepII-tag affinity chromatography. CUL2^sortase-tag•RBX1 eluted from the StrepII affinity column was exchanged to the Sortase Reaction Buffer [50 mM Tris (pH 7.6), 150 mM NaCl, 10 mM CaCl₂] using PD-10 desalting columns (Cytiva), and was then incubated with 60 μM Sortase A enzyme and 250 μM GGGGK^AMCA peptide (synthesized by New England Peptide or as detailed below) at room temperature for 24 hr followed by purification through size exclusion chromatography. The VHL•ELOB•ELOC (VBC) complex was expressed by co-transforming BL21 (DE3) E. Coli cells with the pGEX-4T construct carrying the VHL gene and the pACYCDuet construct carrying both ELOB and ELOC genes. The complex was purified by glutathione affinity chromatography, proteolytic elution by thrombin, and size exclusion chromatography with a Superdex 75 Increase 10/300 GL column (Cytiva). The VHL•ELOB^Lumio-tag•ELOC complex was purified in a similar way and was incubated with FlAsH-EDT₂ (Cayman Chemical) (1:2 molar ratio) at room temperature for two hours in buffer containing 20 mM Tris (pH7.5), 100 mM NaCl, 2 mM TCEP (Thermo Fisher Scientific), 1 mM EDTA, and 5% glycerol. The FlAsH labeled VHL•ELOB^Lumio-tag•ELOC (VBC^FlAsH) was further purified by size exclusion chromatography to remove the excess FlAsH-EDT₂. The ZYG11B•ELOB^Lumio-tag•ELOC complex was purified and labeled similarly as described above for VBC.

The expression and purification of CAND1 or modified CAND1 from bacterial cells followed that described previously³⁰. To generate the baculovirus to express CAND1 in insect cells, the pVL1393 vector carrying the CAND1 gene and the ProGreen linearized Baculorvirus vector DNA (AB Vector) were co-transfected into Sf9 cells using Cellfectin II Reagent (Thermo Fisher Scientific). The resulting baculovirus was further amplified in Sf9 cells and used for large-scale protein expression in Sf21 cells. Recombinant CAND1 expressed in either insect cells or bacterial cells was purified through glutathione-affinity chromatography, digestion with the thrombin protease, anion exchange chromatography, and size exclusion chromatography²⁹. ^GSTCAND1 was expressed and purified in similar ways, omitting the protease cleavage step. ^FlAsHCAND1 was generated by incubating ^Lumio-tagCAND1 with FlAsH-EDT₂ as described above.

Synthesis and purification of GGGGK^AMCA peptide

About 1.2 mg (0.003 mmol) of a peptide consisting of Gly-Gly-Gly-Gly-LysN3 (LysN3, 6-azido-l-lysine) (synthesized by Biomatik) was dissolved in 0.5 mL water with 1 mg (0.004 mmol) of AMCA-alkyne (AAT Bioquest). The “click-chemistry” reaction was initiated by the addition of sodium ascorbate (0.002 mmol) and copper sulfate (0.0002 mmol) which was left to incubate at room temperature overnight. The product GGGGK^AMCA was purified by preparative HPLC over a C18 column with a gradient of water with 0.1% formic acid to 45% acetonitrile. Fractions containing the product (judged by absorbance at 350 nm) were pooled and lyophilized yielding a light-yellow powder.

Degradation of cellular proteins

The assay for CODD degradation was performed as previously described⁴⁸. For PROTAC-induced degradation of bromodomain (BRD) proteins, 6×10⁶ cells of desired genetic background were seeded on 6-well plates one day before the assay. At the time of seeding, expression of genes integrated using the Flp-In system was induced with 1 μg/ml tetracycline, except that no tetracycline was added to the WT^VHL cells (Extended Data Fig. 1). The next day, the PROTAC compounds were added in the cell culture media to initiate targeted protein degradation. When needed, MLN4924 or CSN5i-3 was added to the cell culture media one hour before the addition of PROTACs. At different time points post the addition of PROTACs, cells were washed with PBS and lysed with 2×SDS sample buffer directly in the plate. Cell lysates were then transferred to 1.5 ml tubes and sonicated before SDS-PAGE and Western blot analyses. Fluorescent signals from the secondary antibodies were detected on an Odyssey Imager (LI-COR Biosciences). The intensities of BRD2 and BRD3 signals were normalized to the intensities of the GAPDH signals in the corresponding sample and were fit to a single exponential decay model in GraphPad Prism 9 (GraphPad software) to obtain the half-lives. The degradation assay of ^3xFLAGGFP fused with the wide-type or mutant C-degron of SEPHS2 were similarly performed except that cycloheximide was added to the corresponding cell culture for indicated time periods. To monitor MIC19 degradation, WT and DKO cells were transfected with the vector carrying the ^FLAG-HACHCHD3 gene using Lipofectamine 3000 (Invitrogen). Twenty-four hours after transfection, the cells were seeded into 6-well plates and ^FLAG-HAMIC19 expression was induced as described above. The next day, cells were treated with 60 μg/ml cycloheximide for indicated time periods and analyzed by Western blot. The intensity of FLAG was normalized to that of the GAPDH and then was fit to a single exponential decay model in Prism to obtain the half-lives.

Dissociation of CUL2•VBC in human cell lysate

DKO cells expressing endogenous ^StrepIICUL2 were treated with 1 μM MLN4924 or 1 μM CSN5i-3 for one hour. Cells were then collected and resuspended in the Cell Lysis Buffer [50 mM HEPES (pH 7.5), 5 mM Mg(OAc)₂, 70 mM KOAc, 1 mM DTT, 1 × protease inhibitor cocktail (Roche)] containing 1 μM MLN4924 or 1 μM CSN5i-3 as per cell treatment. After brief sonication, clear cell lysates were collected after centrifugation at 18,000 × g, aliquoted, and incubated with Strep-Tactin resin (IBA Lifesciences). Excess recombinant VHL^114-213•ELOB•ELOC chase protein was added to the cell lysate [~100× of the endogenous VHL level (Extended Data Fig. 3f)] at different time points to initiate the dissociation assay. All sample were incubated with the resin for 120 min at room temperature, and the resin was washed with the Cell Lysis Buffer three times. Proteins bound on the resin were eluted in 2 × SDS sample buffer and analyzed by SDS-PAGE and Western blot. The intensities of endogenous VHL signals were first normalized to the intensities of the StrepII signals in the corresponding sample and then normalized to the sample without chase. The resultant data were fit to a single exponential decay model in Prism to obtain the half-lives. All other cellular immunoprecipitation assays were performed in the same buffer unless otherwise indicated.

Dissociation of CUL2•CAND1 in human cell lysate

DKO cells expressing endogenous ^StrepIICUL2 and tetracycline induced CAND1^HA were treated with 1 μM MLN4924 for one hour. Cells were collected and the dissociation assay was performed as described above, except that recombinant ^GSTCAND1 purified from bacterial cells was used as the chase protein.

Estimation of cellular VHL concentration

Cell lysates of WT HEK293 cells along with recombinant VHL protein samples were immunoblotted using the anti-VHL antibody. A standard curve was generated by fitting VHL signals versus the concentration gradient to a linear-regression model. The dilution fold of VHL by the Cell Lysis Buffer was calculated based on the cell number and the estimation that the volume of a cell approximates a sphere with a diameter of 15 μm.

Dissociation of CUL2 from CAND1 expressed in different cells

^GSTCAND1 proteins purified from insect or bacterial cells were incubated with ^StrepIICUL2 in the Storage Buffer [30 mM Tris (pH 7.5), 100 mM NaCl, 10% glycerol, and 1 mM DTT] followed by immobilization on GST beads for 30 min. During pulldown, excess CUL2^split chase proteins were added to the sample for indicated time. Proteins bound on the beads were washed three times with the Storage Buffer, eluted in 2 × SDS sample buffer and analyzed by SDS-PAGE and Western blot. The intensities of CUL2 signals were first normalized to the intensities of CAND1 signals in the corresponding sample and then normalized to the sample without chase. The resultant data were fit to a single exponential decay model in Prism to obtain the half-lives. All steps were performed at room temperature. All other pulldown assays were performed using the same buffer unless otherwise indicated.

Pulldown assay for CUL2•CAND1 dissociation

CAND1 proteins purified from insect cells were incubated with ^StrepIICUL2 in the Storage Buffer followed by immobilization on Strep-Tactin resin for 120 min at room temperature. During pulldown, excess ^GSTCAND1 chase proteins, with or without VBC, were added to the sample for indicated time. Proteins bound on the beads were washed three times, eluted in 2 × SDS sample buffer and analyzed by SDS-PAGE and Western blot. The intensities of CAND1 signals were first normalized to the intensities of the StrepII signals in the corresponding sample and then normalized to the sample with 5-min chase. The resultant data were fit to a single exponential decay model in Prism to obtain the half-lives.

To determine the effect of VBC and/or neddylation on CUL2•CAND1 dissociation, preassembled 20 nM ^StrepIICUL2•CAND1 complex was incubated with the neddylation reaction mix [50 nM NAE, 600 nM UBC12, 200 nM DCN1 and 400 nM NEDD8] and ^GSTCAND1 chase in the Neddylation Reaction Buffer [30 mM Tris (pH 7.5), 5 mM MgCl₂, 2 mM ATP]. NAE was opted out for no neddylation samples, and 40 nM VBC was added for + VBC samples. After 15-min incubation, all samples were affinity precipitated with the Strep-Tactin resin for 15 min. The bound proteins were washed with the same buffer 3 times and eluted with 2 × SDS sample buffer for Western blot analysis. All steps were performed at room temperature.

Biolayer interferometry assay

The association and dissociation rate constants of the CUL2•VBC complex was measured on an Octet RED384 system (ForteBio) in PBS buffer containing 0.02% Tween 20, 0.1% BSA and 0.05% sodium azide. ^GSTVHL•ELOB•ELOC (^GSTVBC) was immobilized on anti-GST Biosensors (ForteBio) at a concentration of 100 nM to obtain a response value > 1 nm (Extended Data Figs. 3a,b and 4b). The association of ^GSTVBC•CUL2 at various CUL2 concentrations were monitored for 5 min followed by monitoring the complex dissociation for 15 min. Data analysis was performed in Prism. For the association phase, the k_obs at different CUL2 concentrations were obtained by fitting the data to a one-phase association model (Prism). The slope obtained from the regression of k_obs vs. [CUL2] gave the k_on. The k_off was determined by fitting the dissociation data of the highest [CUL2] to a one-phase decay model (Prism).

FRET-based assays

The assays were designed and performed using the methodology described previously⁶⁸. All fluorescent scans were performed on a MOS-200/M fast UV/Vis spectrometer coupled with a MOS-EMM-MONO motorized emission monochromator (BioLogic) in the FRET buffer [30 mM Tris (pH 7.6), 100 mM NaCl, 0.5 mM DTT, and 1 mg/ml Ovalbumin (Sigma)] containing 100 nM of each indicated fluorescent protein. All data were acquired via Bio-Kine32 (BioLogic). For assays using CUL2^AMCA as the FRET donor, all samples were excited at 350 nm and the emission signals were collected from 400 nm to 600 nm with 1 nm increment. For assays using ^CFPCUL2 as the FRET donor, all samples were excited at 430 nm and the emission signals were collected from 450 nm to 650 nm with 1 nm increment. Stopped-flow measurements were performed on an SFM-3000 stopped-flow mixer coupled with a MOS-200/M fast UV/Vis spectrometer (BioLogic) in the FRET buffer. A band-pass filter ET445/58m (Chroma) was used to monitor AMCA signals, and a band-pass filter AT480/30m (Chroma) was used to monitor CFP signals. To determine the k_off of CUL2•VBC, 25 nM ^CFPCUL2 and 25 nM VBC^FlAsH were mixed in the FRET buffer, and aliquots of this mixture were incubated with 250 nM unlabeled VBC chase for desired time periods at room temperature. All samples (150 μl each) were transferred to a black 96-well plate (COSTAR), and the intensity of CFP was measured by a Clariostar microplate reader (BMG Labtech, Cary, NC). Data were normalized to the blank and the sample with no chase protein, and the normalized data were fit into a one-phase decay model in Prism. Three independent replicates were included to obtain the k_off.

To quantify the dissociation rate of CUL2•CAND1, 100 nM CUL2^AMCA and 100 nM ^FlAsHCAND1 were mixed in the FRET buffer. Unlabeled CAND1 chase (1 μM) was added to the mixture. Samples were excited at 350 nm and the peak AMCA signals were collected at indicated time points. Normalized data from three independent replicates were fit into a one-phase decay model in Prism to obtain the k_off.

To examine effects of neddylation on CUL2•CAND1 dissociation, neddylation reaction mix [0.25 μM NAE, 3 μM UBC12, 1 μM DCN1, and 2 μM NEDD8], excess CAND1 chase, and 200 nM VBC, were added to the pre-assembled CUL2^AMCA•^FlAsHCAND1 complex (100 nM each) in the Neddylation Reaction Buffer. NAE was opted out for the no neddylation group. AMCA emission data collection and analyses followed that described above.

SILAC sample preparation

SILAC medium was prepared in Arg- and Lys- depleted DMEM medium (Thermo Scientific) containing 10% dialyzed fetal bovine serum (Cytiva) supplemented with 87 mg/L [¹³C₆, ¹⁵N₄] L-arginine (Arg10) (Sigma-Aldrich) and 153 mg/L [¹³C6, ¹⁵N₂] L-lysine (Lys8) (Sigma-Aldrich), or the corresponding unlabeled Arg and Lys (Sigma-Aldrich). DKO^StrepII-CUL2 HEK293 cells were grown in light SILAC medium whereas DKO or DKO Flp-In HEK293 cells with CAND1 (DKO^CAND1) were grown in heavy SILAC medium (containing Arg10 and Lys8) for 12 days. Cells were treated with either 1 μM MLN4924 or 1 μM CSN5i-3 for one hour and were then collected. Equal number (4x10⁷) of cells from light and heavy SILAC media were lysed in the Cell Lysis Buffer, mixed, and incubated with 100 μl Strep-Tactin resin for 60 min at room temperature. Subsequently, the resin was pelleted via centrifugation at 1,000 × g for 2 min, washed twice with the Cell Lysis Buffer, and washed twice with 100 mM Tris-HCl (pH 8.0). The resin was transferred to a spin column (Bio-Rad, 7326204) and eluted twice through incubating with 150 μl elution buffer [100 mM Tris-HCl (pH 8.0), 50 mM biotin] for 5 min each time. The eluates were buffer exchanged to 100 μl 4 M urea in 10 mM Tris-HCl (pH 8.0) using the Zeba^™ Spin Desalting Columns (Thermo Scientific). At the room temperature, samples were further treated by 5 mM DTT for 30 min (for reduction) and by 10 mM iodoacetamide for 20 min in the dark (for alkylation of cysteine residues). For protease digestion of the eluted proteins, samples were incubated with 1 μg LysC (Wako Chemicals, 125–05061) for 3 h at room temperature, diluted with 50 mM ammonium bicarbonate to bring the concentration of urea to 2 M, and incubated with 1 μg trypsin (Sigma-Aldrich) overnight at 37 °C. The resulting peptides were loaded onto C18 silica micro-spin columns (The Nest Group, SS18V) pre-washed with 100% acetonitrile and pre-equilibrated in Buffer A [0.1% formic acid in water], washed with Buffer A twice and then Buffer B (3% acetonitrile, 97% H₂O and 0.1% formic acid) twice, and eluted with 80% acetonitrile. Small aliquots of the eluates were used to determine the peptide concentration with a Pierce Quantitative Colorimetric Peptide Assay Kit (Thermo Scientific). The rest of eluates were brought to complete dryness in a SpeedVac vacuum concentrator and stored at −80 °C until analyses by mass spectrometry.

Mass spectrometry data collection

Samples were loaded to a 25 cm C18 reverse-phase analytical column (ThermoFisher EasySpray ES902A) with 100% Buffer A using the Easy-nLC 1200 system. Peptides were eluted over a 120-min gradient at a flow rate of 300 nL/min, using Buffer B (80% acetonitrile, 0.1% formic acid) going from 4% to 30% over 90 min, to 50% over 10 min, then to 85% over 10 min, holding at 85% for 5 min, and decreasing to 4% for 5 min.

Spectra were acquired with an Orbitrap Exploris^™ 480 Mass Spectrometer (ThermoFisher Scientific), operating in DD-MS2 mode, with FAIMS Pro^™ Interface (ThermoFisher Scientific) cycling between CVs of −40 V, −55 V and −70 V every 1.3 s. MS1 spectra were acquired at 120,000 resolution with a scan range from 375 to 1500 m/z, standard AGC and auto maximum injection time. HCD spectra were acquired in the ion trap using a normalized collision energy of 30%, an orbitrap resolution of 15000, an isolation window of 1.6 m/z, standard AGC and auto maximum injection time. The precursor intensity threshold was 5e3, charge state was 2-7, and the dynamic exclusion duration was 30 s.

Mass spectrometry data analysis

Raw data were processed in Proteome Discoverer 2.5 (Thermo Fisher Scientific) with a label-free quantification workflow and a customized SILAC 2plex (Arg10, Lys8) quantification method. Minora Feature Detector was used with minimal trace length of 5 min, maximal delta RT of isotope patterns of 0.2 min, minimal S/N threshold of 1, and high PSM confidence. MS/MS spectra were searched against a Uniprot Homo sapiens protein database plus common contaminants (downloaded 100419, 20417 sequences). Search parameters included a maximum of three missed trypsin cleavage events, a precursor ion tolerance of 10 ppm, and a fragment ion tolerance of 0.02 Da. Searches permitted variable modifications of oxidation (M), Label +8.014 Da (K), Label +10.008 Da (R), GG (K), deamidation (N,Q); protein N-terminus dynamic modifications of Acetyl, Met-loss, and Met-loss plus Acetyl, and static modifications of cysteine (+57.0215 Da) for carbamidomethylation. The confidence of peptide discovery was estimated with the percolator node based on the q value and decoy database search, using a strict target FDR of 0.01 and a relaxed target FDR of 0.05. Precursor Ion quantification was performed using a feature mapper node with RT alignment, a max RT shift of 10 min as well as mass tolerances of 10 ppm. Both unique and razor peptides were used for precursor ion quantification based on the peak intensities. Protein abundances were calculated using summed peptide abundances. The Protein Validator node was used to annotate the data at a maximum of 1% FDR. SILAC H/L ratios of proteins were calculated using the median of all pairwise peptide ratios.

Statistics and reproducibility

Western blot signals were quantified by Image Studio Lite (LI-COR Biosciences). Statistical analyses were performed in GraphPad Prism. All p values were determined by two-sided t-tests with no adjustments. Details of the statistical analyses are provided in figures and figure legends. Data shown in Figs. 1a,1b, 3c, 4c, 6h, 6i, Extended Data Figs. 3e, 3g, 5c, and 7a-c were representative results from two independent experiments.

Extended Data

Extended Data Fig. 3 — a, Trial of bait protein immobilization for the BLI experiments. ^GSTVBC at various concentrations were incubated with anti-GST antibody coated biosensors, and the sensor response levels were recorded over time. For all BLI experiments in this study, 100 nM ^GSTVBC was used for loading, and a minimal response level of 1 nm was achieved before the kinetic binding assay was initiated.

b, Representative results of ^GSTVBC loading for the BLI assay shown in Fig. 3a.

c, Schematic illustration of CRISPR/Cas9-mediated epitope tagging on the endogenous *CUL2* locus in DKO cells. Single-stranded oligodeoxynucleotide (ssODN) carrying the coding sequence for the StrepII tag was used as the homology-directed repair (HDR) template. Red arrows indicate primers (1F, 2F, 3F, 1R, 2R) used for screening the correctly tagged cells.

d, The guide RNA (gRNA) targeting *CUL2* as illustrated in c is efficient in inducing double strain breaks at the target site. Cell populations electroporated with the plasmid expressing gRNA and Cas9 were harvested for genomic DNA extraction and PCR with the 3F and 1R primer set as illustrated in c. Sequencing chromatograms of PCR products spanning the target sites were shown.

e, Successful gene editing in the knock-in clonal cell line was identified using the primer sets “1F + 1R” and “2F + 2R” as illustrated in c.

f, Sequencing chromatograms confirming the correct genomic sequences at the 5’ and 3’ junctions of the HDR region in the knock-in cell line.

g, Immunoblots validating the successful preparation of SILAC samples illustrated in Fig. 3d.

h, Estimation of the VHL concentration in HEK293 cells. Total cell extract from WT cells and recombinant VHL standards were immunoblotted with anti-VHL antibody. Levels of VHL in the WT cells were calculated based on the standard curve generated from the known amounts of recombinant VHL standards.

Extended Data Fig. 4 — a, Sequence alignment of ELOC^1-40 from different species.

b, Representative results of ^GSTVBC loading for the BLI assay shown in Fig. 4a.

**c,d,** BLI analyses revealing k_on **(c)** and k_off **(d)** for CUL2 binding to VBC that contains different ELOC isoforms. n = 3. See the “Methods” section for more details.

e, Measurements of k_obs for VBC binding to CUL2. Real-time CFP fluorescence upon mixing increasing concentrations of VBC^FlAsH to 10 nM ^CFPCUL2 in a stopped-flow fluorimeter. Signal changes with background (determined by the “buffer” group) subtraction were fit to two-phase exponential curves. The fast-phase and slow-phase rates (k_obs) were plotted to give k_on,fast and k_on,slow, respectively.

**f,g,** Estimation of the CUL2•CAND1 dissociation rate in human cell lysate. f, Schematic workflow of the assay. Relative levels of co-precipitated CAND1^HA were calculated from CAND1^HA/StrepIICUL2 intensity ratios followed by normalizing these ratios to that obtained from the 0-min sample. The normalized data were fit to a single exponential curve to obtain t_1/2 and k_off **(g)**. n = 2.

**h,i,** CAND1 produced from insect cells, but not bacterial cells, forms CUL2•CAND1 as stable as the CUL2•CAND1 in human cell lysate **(f,g,)**. h, GST pulldown analyses of recombinant CUL2 bound to ^GSTCAND1 produced from E. coli or insect cells in the presence of CUL2^split chase for indicated time. Samples were fractionated on SDS-PAGE gels and immunoblotted with indicated antibodies. Intensity ratios of the CUL2 and ^GSTCAND1 bands were fit to single exponential curves to obtain t_1/2 **(i)**. n = 3.

j, Measurement of k_obs for CAND1 binding to CUL2. Real-time AMCA fluorescence upon addition of increasing concentrations of ^FlAsHCAND1 to CUL2^AMCA. Signal changes were fit to one phase exponential curves. CAND1 was purified from insect cells.

Error bars represent SEM (d) or SD (i). “n” indicates number of independent experiments. P values were determined by two-sided t-tests with no adjustments.

Extended Data Fig. 5 — a, Measurements of k_obs for VBC binding to CUL2 preassembled with CAND1. Same as in Extended Data Fig. 4e except that 20 nM CAND1 was preincubated with 10 nM ^CFPCUL2. Signal changes with background subtraction (background determined by the “buffer” group) were fit to one-phase exponential curves to give k_obs. CAND1 was produced from insect cells.

**b,c,** CAND1 accelerates the disassembly of CRL2^VHL in human cell lysate. The schematic workflow of the experiment is depicted in b. Samples were immunoblotted with anti-VHL, anti-StrepII, and anti-CAND1 antibodies **(c)**. Both short (S.E.) and long (L.E.) exposure for anti-VHL signals are shown.

d, CAND1 accelerates the disassembly of CUL2•VBC *in vitro*. Same as in Fig. 5c except with 100 nM of different CAND1 proteins added in the presence of VBC chase. Both CAND1 and CAND1β++ were generated from bacterial cells. CAND1β++ showed no significant effect on the ^CFPCUL2-VBC^FlAsH FRET signal.

e, Structural alignment of CUL1•CAND1 complex (PDB ID: 1U6G) and CUL2•VBC complex (PDB ID: 5N4W) based on the N-terminal amino acid sequences (1-300) of cullins.

f, Measurements of k_obs for CUL2•VBC dissociation by CAND1. Real-time detection of CFP fluorescence in a stopped-flow apparatus upon addition of increasing concentrations of CAND1 together with 100 nM VBC chase to 10 nM ^CFPCUL2 preincubated with 10 nM VBC^FlAsH. Final concentrations after mixing the two solutions in the stopped-flow fluorimeter were shown. Signal changes were fit to single exponential curves to obtain the k_obs. CAND1 was produced from insect cells.

a, Measurements of k_obs for ^FlAsHCAND1 binding to CUL2^AMCA preassembled with VBC. Same as in Extended Data Fig. 4j except that 20 nM VBC was preincubated with 5 nM CUL2^AMCA. Signal changes were fit to one phase exponential curves. CAND1 was purified from insect cells.

**b,c,** k_on for CAND1 binding to CUL2 preassembled with VBC. b, Measurement of k_obs for ^FlAsHCAND1 binding to CUL2^AMCA preassembled with VBC. Same as in Extended Data Fig. 6a except that 1 μM VBC was preincubated with 5 nM CUL2^AMCA. Signal changes were fit to one-phase exponential curves to obtain k_obs, which were plotted against ^FlAsHCAND1 concentrations in **(c)**. Linear slope in **(c)** gave the k_on. Error bars represent SEM; n = 4 independent experiments. CAND1 was purified from insect cells.

d, HEK293 cells were treated with 1 μM CSN5i-3 for indicated time, and the rate of CUL2 neddylation was monitored using immunoblot with anti-CUL2 antibody. The proportion of unneddylated CUL2 in the whole CUL2 population (CUL2^Total) versus time was plotted and fit to a single exponential curve to obtain the k_obs. Error bars represent SEM; n = 3 independent experiments. CUL2^N8: neddylated CUL2.

Extended Data Fig. 7 — a, StrepII-tag affinity pulldown assay confirming the binding between recombinant ^StrePIICUL2 and ZBC.

**b,c,** Similar to Fig. 6h,i except that VBC was replaced by ZBC. Samples were immunoblotted with anti-CAND1, anti-StrepII, and anti-ELOB antibodies.

Extended Data Fig. 8 — a, Schematic illustration for the MZ-1 and SIM1 mediated interactions between VHL and bromodomain proteins.

b, As in Fig. 7a except that 0.3 μM MZ-1 was used in place of 1 μM MZ-1. Error bars represent SEM; n = 3 independent experiments.

ACKNOWLEDGEMENTS

We thank Dr. Shu-Ou Shan (California Institute of Technology) for helpful discussion on kinetics data. We thank Dr. Stephen J. Elledge (Harvard Medical School) and Dr. Daniel C. Scott (St. Jude Children's Research Hospital) for reagents. Mass spectrometry was provided by the Indiana University School of Medicine Center for Proteome Analysis. The Octet RED384 system for BLI assays was provided by the Chemical Genomics Facility at Purdue Institute for Drug Discovery. This work was supported by National Institutes of Health grant R35GM138016 (to X.L.) and American Heart Association Career Development Award (to X.L.). The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.

Footnotes

Publisher's Disclaimer: This version of the article has been accepted for publication, after peer review, but is not the Version of Record and does not reflect post-acceptance improvements, or any corrections. The Version of Record is available online at: http://dx.doi.org/10.1038/s41594-023-01167-5. Use of this Accepted Version is subject to the publisher’s Accepted Manuscript terms of use https://www.springernature.com/gp/open-research/policies/accepted-manuscript-terms.

COMPETING INTERESTS STATEMENT

The authors declare no competing interests.

DATA AVAILABILITY STATEMENT

All data supporting the findings of this study are available within the paper. Source data are provided. The proteomics data have been deposited to the PRIDE with the dataset identifier PXD045609. The Uniprot Homo sapiens protein database is accessible through the link below: https://www.uniprot.org/proteomes/.

REFERENCES

1.Pohl C & Dikic I Cellular quality control by the ubiquitin-proteasome system and autophagy. Science 366, 818–822 (2019). [DOI] [PubMed] [Google Scholar]
2.Harper JW & Schulman BA Cullin-RING Ubiquitin Ligase Regulatory Circuits: A Quarter Century Beyond the F-Box Hypothesis. Annu Rev Biochem 90, 403–429 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Lydeard JR, Schulman BA & Harper JW Building and remodelling Cullin-RING E3 ubiquitin ligases. EMBO Rep 14, 1050–61 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Zimmerman ES, Schulman BA & Zheng N Structural assembly of cullin-RING ubiquitin ligase complexes. Curr Opin Struct Biol 20, 714–21 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Diaz S, Wang K, Sjogren B & Liu X Roles of Cullin-RING Ubiquitin Ligases in Cardiovascular Diseases. Biomolecules 12(2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Kamura T, Conrad MN, Yan Q, Conaway RC & Conaway JW The Rbx1 subunit of SCF and VHL E3 ubiquitin ligase activates Rub1 modification of cullins Cdc53 and Cul2. Genes Dev 13, 2928–33 (1999). [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Kamura T. et al. VHL-box and SOCS-box domains determine binding specificity for Cul2-Rbx1 and Cul5-Rbx2 modules of ubiquitin ligases. Genes Dev 18, 3055–65 (2004). [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Mahrour N. et al. Characterization of Cullin-box sequences that direct recruitment of Cul2-Rbx1 and Cul5-Rbx2 modules to Elongin BC-based ubiquitin ligases. J Biol Chem 283, 8005–13 (2008). [DOI] [PubMed] [Google Scholar]
9.Sherpa D, Chrustowicz J & Schulman BA How the ends signal the end: Regulation by E3 ubiquitin ligases recognizing protein termini. Mol Cell 82, 1424–1438 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Maxwell PH et al. The tumour suppressor protein VHL targets hypoxia-inducible factors for oxygen-dependent proteolysis. Nature 399, 271–5 (1999). [DOI] [PubMed] [Google Scholar]
11.Liu X, Zurlo G & Zhang Q The Roles of Cullin-2 E3 Ubiquitin Ligase Complex in Cancer. Adv Exp Med Biol 1217, 173–186 (2020). [DOI] [PubMed] [Google Scholar]
12.Bekes M, Langley DR & Crews CM PROTAC targeted protein degraders: the past is prologue. Nat Rev Drug Discov 21, 181–200 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Cowan AD & Ciulli A Driving E3 Ligase Substrate Specificity for Targeted Protein Degradation: Lessons from Nature and the Laboratory. Annu Rev Biochem 91, 295–319 (2022). [DOI] [PubMed] [Google Scholar]
14.Baek K, Scott DC & Schulman BA NEDD8 and ubiquitin ligation by cullin-RING E3 ligases. Curr Opin Struct Biol 67, 101–109 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Baek K. et al. NEDD8 nucleates a multivalent cullin-RING-UBE2D ubiquitin ligation assembly. Nature 578, 461–466 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Deshaies RJ & Pierce NW Transfer of ubiquitin protein caught in the act. Nature 578, 372–373 (2020). [DOI] [PubMed] [Google Scholar]
17.Saha A & Deshaies RJ Multimodal activation of the ubiquitin ligase SCF by Nedd8 conjugation. Mol Cell 32, 21–31 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Rozen S. et al. CSNAP Is a Stoichiometric Subunit of the COP9 Signalosome. Cell Rep 13, 585–598 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Gutierrez C. et al. Structural dynamics of the human COP9 signalosome revealed by cross-linking mass spectrometry and integrative modeling. Proc Natl Acad Sci U S A 117, 4088–4098 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Schulze-Niemand E & Naumann M The COP9 signalosome: A versatile regulatory hub of Cullin-RING ligases. Trends Biochem Sci (2022). [DOI] [PubMed] [Google Scholar]
21.Zhang Y. et al. Adaptive exchange sustains cullin-RING ubiquitin ligase networks and proper licensing of DNA replication. Proc Natl Acad Sci U S A 119, e2205608119 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Pintard L. et al. Neddylation and Deneddylation of CUL-3 Is Required to Target MEI-1/Katanin for Degradation at the Meiosis-to-Mitosis Transition in C. elegans. Current Biology 13, 911–921 (2003). [DOI] [PubMed] [Google Scholar]
23.Wolf DA, Zhou C & Wee S The COP9 signalosome: an assembly and maintenance platform for cullin ubiquitin ligases? Nat Cell Biol 5, 1029–33 (2003). [DOI] [PubMed] [Google Scholar]
24.Cope GA & Deshaies RJ COP9 signalosome: a multifunctional regulator of SCF and other cullin-based ubiquitin ligases. Cell 114, 663–71 (2003). [DOI] [PubMed] [Google Scholar]
25.Zheng J. et al. CAND1 Binds to Unneddylated CUL1 and Regulates the Formation of SCF Ubiquitin E3 Ligase Complex. Molecular Cell 10, 1519–1526 (2002). [DOI] [PubMed] [Google Scholar]
26.Liu J, Furukawa M, Matsumoto T & Xiong Y NEDD8 Modification of CUL1 Dissociates p120CAND1, an Inhibitor of CUL1-SKP1 Binding and SCF Ligases. Molecular Cell 10, 1511–1518 (2002). [DOI] [PubMed] [Google Scholar]
27.Goldenberg SJ et al. Structure of the Cand1-Cul1-Roc1 complex reveals regulatory mechanisms for the assembly of the multisubunit cullin-dependent ubiquitin ligases. Cell 119, 517–28 (2004). [DOI] [PubMed] [Google Scholar]
28.Reitsma JM et al. Composition and Regulation of the Cellular Repertoire of SCF Ubiquitin Ligases. Cell 171, 1326–1339 e14 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Liu X. et al. Cand1-Mediated Adaptive Exchange Mechanism Enables Variation in F-Box Protein Expression. Mol Cell 69, 773–786 e6 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Pierce NW et al. Cand1 Promotes Assembly of New SCF Complexes through Dynamic Exchange of F Box Proteins. Cell 153, 206–15 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Baek K. et al. Systemwide disassembly and assembly of SCF ubiquitin ligase complexes. Cell 186, 1895–1911 e21 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Wang K, Deshaies RJ & Liu X Assembly and Regulation of CRL Ubiquitin Ligases. Adv Exp Med Biol 1217, 33–46 (2020). [DOI] [PubMed] [Google Scholar]
33.Li L, Wang K, Zhou Y & Liu X Review: A silent concert in developing plants: Dynamic assembly of cullin-RING ubiquitin ligases. Plant Sci 330, 111662 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Li L. et al. CAND1 is required for pollen viability in Arabidopsis thaliana-a test of the adaptive exchange hypothesis. Front Plant Sci 13, 866086 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Wu S. et al. CAND1 controls in vivo dynamics of the cullin 1-RING ubiquitin ligase repertoire. Nat Commun 4, 1642 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
36.Zemla A. et al. CSN- and CAND1-dependent remodelling of the budding yeast SCF complex. Nat Commun 4, 1641 (2013). [DOI] [PubMed] [Google Scholar]
37.Zhang W. et al. Genetic analysis of CAND1-CUL1 interactions in Arabidopsis supports a role for CAND1-mediated cycling of the SCF^TIR1 complex. Proc Natl Acad Sci U S A 105, 8470–5 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]
38.Chuang HW, Zhang W & Gray WM Arabidopsis ETA2, an apparent ortholog of the human cullin-interacting protein CAND1, is required for auxin responses mediated by the SCF(TIR1) ubiquitin ligase. Plant Cell 16, 1883–97 (2004). [DOI] [PMC free article] [PubMed] [Google Scholar]
39.Feng S. et al. Arabidopsis CAND1, an unmodified CUL1-interacting protein, is involved in multiple developmental pathways controlled by ubiquitin/proteasome-mediated protein Degradation. Plant Cell 16, 1870–82 (2004). [DOI] [PMC free article] [PubMed] [Google Scholar]
40.Rusnac DV & Zheng N Structural Biology of CRL Ubiquitin Ligases. Adv Exp Med Biol 1217, 9–31 (2020). [DOI] [PubMed] [Google Scholar]
41.Lo SC & Hannink M CAND1-mediated substrate adaptor recycling is required for efficient repression of Nrf2 by Keap1. Mol Cell Biol 26, 1235–44 (2006). [DOI] [PMC free article] [PubMed] [Google Scholar]
42.Reichermeier KM et al. PIKES Analysis Reveals Response to Degraders and Key Regulatory Mechanisms of the CRL4 Network. Mol Cell 77, 1092–1106 e9 (2020). [DOI] [PubMed] [Google Scholar]
43.Mayor-Ruiz C. et al. Plasticity of the Cullin-RING Ligase Repertoire Shapes Sensitivity to Ligand-Induced Protein Degradation. Mol Cell 75, 849–858 e8 (2019). [DOI] [PubMed] [Google Scholar]
44.Sievers QL, Gasser JA, Cowley GS, Fischer ES & Ebert BL Genome-wide screen identifies cullin-RING ligase machinery required for lenalidomide-dependent CRL4(CRBN) activity. Blood 132, 1293–1303 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
45.Soucy TA et al. An inhibitor of NEDD8-activating enzyme as a new approach to treat cancer. Nature 458, 732–6 (2009). [DOI] [PubMed] [Google Scholar]
46.Diaz S, Li L, Wang K & Liu X Expression and purification of functional recombinant CUL2*RBX1 from E. coli. Sci Rep 11, 11224 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
47.Raina K. et al. PROTAC-induced BET protein degradation as a therapy for castration-resistant prostate cancer. Proc Natl Acad Sci U S A 113, 7124–9 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
48.Wang K, Reichermeier KM & Liu X Quantitative analyses for effects of neddylation on CRL2(VHL) substrate ubiquitination and degradation. Protein Sci 30, 2338–2345 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
49.Koren I. et al. The Eukaryotic Proteome Is Shaped by E3 Ubiquitin Ligases Targeting C-Terminal Degrons. Cell 173, 1622–1635 e14 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
50.Yeh CW et al. The C-degron pathway eliminates mislocalized proteins and products of deubiquitinating enzymes. EMBO J 40, e105846 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
51.Schlierf A. et al. Targeted inhibition of the COP9 signalosome for treatment of cancer. Nat Commun 7, 13166 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
52.Cope GA et al. Role of predicted metalloprotease motif of Jab1/Csn5 in cleavage of Nedd8 from Cul1. Science 298, 608–11 (2002). [DOI] [PubMed] [Google Scholar]
53.Schwechheimer C. et al. Interactions of the COP9 signalosome with the E3 ubiquitin ligase SCFTIRI in mediating auxin response. Science 292, 1379–82 (2001). [DOI] [PubMed] [Google Scholar]
54.Schulze-Niemand E & Naumann M The COP9 signalosome: A versatile regulatory hub of Cullin-RING ligases. Trends Biochem Sci 48, 82–95 (2023). [DOI] [PubMed] [Google Scholar]
55.Nguyen HC, Yang H, Fribourgh JL, Wolfe LS & Xiong Y Insights into Cullin-RING E3 ubiquitin ligase recruitment: structure of the VHL-EloBC-Cul2 complex. Structure 23, 441–449 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
56.Cardote TAF, Gadd MS & Ciulli A Crystal Structure of the Cul2-Rbx1-EloBC-VHL Ubiquitin Ligase Complex. Structure 25, 901–911 e3 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
57.Mosadeghi R. et al. Structural and kinetic analysis of the COP9-Signalosome activation and the cullin-RING ubiquitin ligase deneddylation cycle. Elife 5(2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
58.Faull SV et al. Structural basis of Cullin 2 RING E3 ligase regulation by the COP9 signalosome. Nat Commun 10, 3814 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
59.Shaaban M. et al. Structural and mechanistic insights into the CAND1-mediated SCF substrate receptor exchange. Mol Cell 83, 2332–2346 e8 (2023). [DOI] [PubMed] [Google Scholar]
60.Roy MJ et al. SPR-Measured Dissociation Kinetics of PROTAC Ternary Complexes Influence Target Degradation Rate. ACS Chem Biol 14, 361–368 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
61.Gadd MS et al. Structural basis of PROTAC cooperative recognition for selective protein degradation. Nat Chem Biol 13, 514–521 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
62.Imaide S. et al. Trivalent PROTACs enhance protein degradation via combined avidity and cooperativity. Nat Chem Biol 17, 1157–1167 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
63.Lin HC et al. CRL2 aids elimination of truncated selenoproteins produced by failed UGA/Sec decoding. Science 349, 91–5 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
64.Lin HC et al. C-Terminal End-Directed Protein Elimination by CRL2 Ubiquitin Ligases. Mol Cell 70, 602–613 e3 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
65.Timms RT et al. A glycine-specific N-degron pathway mediates the quality control of protein N-myristoylation. Science 365(2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
66.Fischer ES et al. The molecular basis of CRL4DDB2/CSA ubiquitin ligase architecture, targeting, and activation. Cell 147, 1024–39 (2011). [DOI] [PubMed] [Google Scholar]

METHODS-ONLY REFERENCES

67.Wang K & Liu X Determining the Effects of Neddylation on Cullin-RING Ligase-Dependent Protein Ubiquitination. Curr Protoc 2, e401 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
68.Garsamo M, Zhou Y & Liu X Using In Vitro Fluorescence Resonance Energy Transfer to Study the Dynamics Of Protein Complexes at a Millisecond Time Scale. J Vis Exp (2019). [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

[R1] 1.Pohl C & Dikic I Cellular quality control by the ubiquitin-proteasome system and autophagy. Science 366, 818–822 (2019). [DOI] [PubMed] [Google Scholar]

[R2] 2.Harper JW & Schulman BA Cullin-RING Ubiquitin Ligase Regulatory Circuits: A Quarter Century Beyond the F-Box Hypothesis. Annu Rev Biochem 90, 403–429 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R3] 3.Lydeard JR, Schulman BA & Harper JW Building and remodelling Cullin-RING E3 ubiquitin ligases. EMBO Rep 14, 1050–61 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R4] 4.Zimmerman ES, Schulman BA & Zheng N Structural assembly of cullin-RING ubiquitin ligase complexes. Curr Opin Struct Biol 20, 714–21 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R5] 5.Diaz S, Wang K, Sjogren B & Liu X Roles of Cullin-RING Ubiquitin Ligases in Cardiovascular Diseases. Biomolecules 12(2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R6] 6.Kamura T, Conrad MN, Yan Q, Conaway RC & Conaway JW The Rbx1 subunit of SCF and VHL E3 ubiquitin ligase activates Rub1 modification of cullins Cdc53 and Cul2. Genes Dev 13, 2928–33 (1999). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R7] 7.Kamura T. et al. VHL-box and SOCS-box domains determine binding specificity for Cul2-Rbx1 and Cul5-Rbx2 modules of ubiquitin ligases. Genes Dev 18, 3055–65 (2004). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R8] 8.Mahrour N. et al. Characterization of Cullin-box sequences that direct recruitment of Cul2-Rbx1 and Cul5-Rbx2 modules to Elongin BC-based ubiquitin ligases. J Biol Chem 283, 8005–13 (2008). [DOI] [PubMed] [Google Scholar]

[R9] 9.Sherpa D, Chrustowicz J & Schulman BA How the ends signal the end: Regulation by E3 ubiquitin ligases recognizing protein termini. Mol Cell 82, 1424–1438 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R10] 10.Maxwell PH et al. The tumour suppressor protein VHL targets hypoxia-inducible factors for oxygen-dependent proteolysis. Nature 399, 271–5 (1999). [DOI] [PubMed] [Google Scholar]

[R11] 11.Liu X, Zurlo G & Zhang Q The Roles of Cullin-2 E3 Ubiquitin Ligase Complex in Cancer. Adv Exp Med Biol 1217, 173–186 (2020). [DOI] [PubMed] [Google Scholar]

[R12] 12.Bekes M, Langley DR & Crews CM PROTAC targeted protein degraders: the past is prologue. Nat Rev Drug Discov 21, 181–200 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R13] 13.Cowan AD & Ciulli A Driving E3 Ligase Substrate Specificity for Targeted Protein Degradation: Lessons from Nature and the Laboratory. Annu Rev Biochem 91, 295–319 (2022). [DOI] [PubMed] [Google Scholar]

[R14] 14.Baek K, Scott DC & Schulman BA NEDD8 and ubiquitin ligation by cullin-RING E3 ligases. Curr Opin Struct Biol 67, 101–109 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R15] 15.Baek K. et al. NEDD8 nucleates a multivalent cullin-RING-UBE2D ubiquitin ligation assembly. Nature 578, 461–466 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R16] 16.Deshaies RJ & Pierce NW Transfer of ubiquitin protein caught in the act. Nature 578, 372–373 (2020). [DOI] [PubMed] [Google Scholar]

[R17] 17.Saha A & Deshaies RJ Multimodal activation of the ubiquitin ligase SCF by Nedd8 conjugation. Mol Cell 32, 21–31 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R18] 18.Rozen S. et al. CSNAP Is a Stoichiometric Subunit of the COP9 Signalosome. Cell Rep 13, 585–598 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R19] 19.Gutierrez C. et al. Structural dynamics of the human COP9 signalosome revealed by cross-linking mass spectrometry and integrative modeling. Proc Natl Acad Sci U S A 117, 4088–4098 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R20] 20.Schulze-Niemand E & Naumann M The COP9 signalosome: A versatile regulatory hub of Cullin-RING ligases. Trends Biochem Sci (2022). [DOI] [PubMed] [Google Scholar]

[R21] 21.Zhang Y. et al. Adaptive exchange sustains cullin-RING ubiquitin ligase networks and proper licensing of DNA replication. Proc Natl Acad Sci U S A 119, e2205608119 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R22] 22.Pintard L. et al. Neddylation and Deneddylation of CUL-3 Is Required to Target MEI-1/Katanin for Degradation at the Meiosis-to-Mitosis Transition in C. elegans. Current Biology 13, 911–921 (2003). [DOI] [PubMed] [Google Scholar]

[R23] 23.Wolf DA, Zhou C & Wee S The COP9 signalosome: an assembly and maintenance platform for cullin ubiquitin ligases? Nat Cell Biol 5, 1029–33 (2003). [DOI] [PubMed] [Google Scholar]

[R24] 24.Cope GA & Deshaies RJ COP9 signalosome: a multifunctional regulator of SCF and other cullin-based ubiquitin ligases. Cell 114, 663–71 (2003). [DOI] [PubMed] [Google Scholar]

[R25] 25.Zheng J. et al. CAND1 Binds to Unneddylated CUL1 and Regulates the Formation of SCF Ubiquitin E3 Ligase Complex. Molecular Cell 10, 1519–1526 (2002). [DOI] [PubMed] [Google Scholar]

[R26] 26.Liu J, Furukawa M, Matsumoto T & Xiong Y NEDD8 Modification of CUL1 Dissociates p120CAND1, an Inhibitor of CUL1-SKP1 Binding and SCF Ligases. Molecular Cell 10, 1511–1518 (2002). [DOI] [PubMed] [Google Scholar]

[R27] 27.Goldenberg SJ et al. Structure of the Cand1-Cul1-Roc1 complex reveals regulatory mechanisms for the assembly of the multisubunit cullin-dependent ubiquitin ligases. Cell 119, 517–28 (2004). [DOI] [PubMed] [Google Scholar]

[R28] 28.Reitsma JM et al. Composition and Regulation of the Cellular Repertoire of SCF Ubiquitin Ligases. Cell 171, 1326–1339 e14 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R29] 29.Liu X. et al. Cand1-Mediated Adaptive Exchange Mechanism Enables Variation in F-Box Protein Expression. Mol Cell 69, 773–786 e6 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R30] 30.Pierce NW et al. Cand1 Promotes Assembly of New SCF Complexes through Dynamic Exchange of F Box Proteins. Cell 153, 206–15 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R31] 31.Baek K. et al. Systemwide disassembly and assembly of SCF ubiquitin ligase complexes. Cell 186, 1895–1911 e21 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R32] 32.Wang K, Deshaies RJ & Liu X Assembly and Regulation of CRL Ubiquitin Ligases. Adv Exp Med Biol 1217, 33–46 (2020). [DOI] [PubMed] [Google Scholar]

[R33] 33.Li L, Wang K, Zhou Y & Liu X Review: A silent concert in developing plants: Dynamic assembly of cullin-RING ubiquitin ligases. Plant Sci 330, 111662 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R34] 34.Li L. et al. CAND1 is required for pollen viability in Arabidopsis thaliana-a test of the adaptive exchange hypothesis. Front Plant Sci 13, 866086 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R35] 35.Wu S. et al. CAND1 controls in vivo dynamics of the cullin 1-RING ubiquitin ligase repertoire. Nat Commun 4, 1642 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R36] 36.Zemla A. et al. CSN- and CAND1-dependent remodelling of the budding yeast SCF complex. Nat Commun 4, 1641 (2013). [DOI] [PubMed] [Google Scholar]

[R37] 37.Zhang W. et al. Genetic analysis of CAND1-CUL1 interactions in Arabidopsis supports a role for CAND1-mediated cycling of the SCF^TIR1 complex. Proc Natl Acad Sci U S A 105, 8470–5 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R38] 38.Chuang HW, Zhang W & Gray WM Arabidopsis ETA2, an apparent ortholog of the human cullin-interacting protein CAND1, is required for auxin responses mediated by the SCF(TIR1) ubiquitin ligase. Plant Cell 16, 1883–97 (2004). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R39] 39.Feng S. et al. Arabidopsis CAND1, an unmodified CUL1-interacting protein, is involved in multiple developmental pathways controlled by ubiquitin/proteasome-mediated protein Degradation. Plant Cell 16, 1870–82 (2004). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R40] 40.Rusnac DV & Zheng N Structural Biology of CRL Ubiquitin Ligases. Adv Exp Med Biol 1217, 9–31 (2020). [DOI] [PubMed] [Google Scholar]

[R41] 41.Lo SC & Hannink M CAND1-mediated substrate adaptor recycling is required for efficient repression of Nrf2 by Keap1. Mol Cell Biol 26, 1235–44 (2006). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R42] 42.Reichermeier KM et al. PIKES Analysis Reveals Response to Degraders and Key Regulatory Mechanisms of the CRL4 Network. Mol Cell 77, 1092–1106 e9 (2020). [DOI] [PubMed] [Google Scholar]

[R43] 43.Mayor-Ruiz C. et al. Plasticity of the Cullin-RING Ligase Repertoire Shapes Sensitivity to Ligand-Induced Protein Degradation. Mol Cell 75, 849–858 e8 (2019). [DOI] [PubMed] [Google Scholar]

[R44] 44.Sievers QL, Gasser JA, Cowley GS, Fischer ES & Ebert BL Genome-wide screen identifies cullin-RING ligase machinery required for lenalidomide-dependent CRL4(CRBN) activity. Blood 132, 1293–1303 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R45] 45.Soucy TA et al. An inhibitor of NEDD8-activating enzyme as a new approach to treat cancer. Nature 458, 732–6 (2009). [DOI] [PubMed] [Google Scholar]

[R46] 46.Diaz S, Li L, Wang K & Liu X Expression and purification of functional recombinant CUL2*RBX1 from E. coli. Sci Rep 11, 11224 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R47] 47.Raina K. et al. PROTAC-induced BET protein degradation as a therapy for castration-resistant prostate cancer. Proc Natl Acad Sci U S A 113, 7124–9 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R48] 48.Wang K, Reichermeier KM & Liu X Quantitative analyses for effects of neddylation on CRL2(VHL) substrate ubiquitination and degradation. Protein Sci 30, 2338–2345 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R49] 49.Koren I. et al. The Eukaryotic Proteome Is Shaped by E3 Ubiquitin Ligases Targeting C-Terminal Degrons. Cell 173, 1622–1635 e14 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R50] 50.Yeh CW et al. The C-degron pathway eliminates mislocalized proteins and products of deubiquitinating enzymes. EMBO J 40, e105846 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R51] 51.Schlierf A. et al. Targeted inhibition of the COP9 signalosome for treatment of cancer. Nat Commun 7, 13166 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R52] 52.Cope GA et al. Role of predicted metalloprotease motif of Jab1/Csn5 in cleavage of Nedd8 from Cul1. Science 298, 608–11 (2002). [DOI] [PubMed] [Google Scholar]

[R53] 53.Schwechheimer C. et al. Interactions of the COP9 signalosome with the E3 ubiquitin ligase SCFTIRI in mediating auxin response. Science 292, 1379–82 (2001). [DOI] [PubMed] [Google Scholar]

[R54] 54.Schulze-Niemand E & Naumann M The COP9 signalosome: A versatile regulatory hub of Cullin-RING ligases. Trends Biochem Sci 48, 82–95 (2023). [DOI] [PubMed] [Google Scholar]

[R55] 55.Nguyen HC, Yang H, Fribourgh JL, Wolfe LS & Xiong Y Insights into Cullin-RING E3 ubiquitin ligase recruitment: structure of the VHL-EloBC-Cul2 complex. Structure 23, 441–449 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R56] 56.Cardote TAF, Gadd MS & Ciulli A Crystal Structure of the Cul2-Rbx1-EloBC-VHL Ubiquitin Ligase Complex. Structure 25, 901–911 e3 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R57] 57.Mosadeghi R. et al. Structural and kinetic analysis of the COP9-Signalosome activation and the cullin-RING ubiquitin ligase deneddylation cycle. Elife 5(2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R58] 58.Faull SV et al. Structural basis of Cullin 2 RING E3 ligase regulation by the COP9 signalosome. Nat Commun 10, 3814 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R59] 59.Shaaban M. et al. Structural and mechanistic insights into the CAND1-mediated SCF substrate receptor exchange. Mol Cell 83, 2332–2346 e8 (2023). [DOI] [PubMed] [Google Scholar]

[R60] 60.Roy MJ et al. SPR-Measured Dissociation Kinetics of PROTAC Ternary Complexes Influence Target Degradation Rate. ACS Chem Biol 14, 361–368 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R61] 61.Gadd MS et al. Structural basis of PROTAC cooperative recognition for selective protein degradation. Nat Chem Biol 13, 514–521 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R62] 62.Imaide S. et al. Trivalent PROTACs enhance protein degradation via combined avidity and cooperativity. Nat Chem Biol 17, 1157–1167 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R63] 63.Lin HC et al. CRL2 aids elimination of truncated selenoproteins produced by failed UGA/Sec decoding. Science 349, 91–5 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R64] 64.Lin HC et al. C-Terminal End-Directed Protein Elimination by CRL2 Ubiquitin Ligases. Mol Cell 70, 602–613 e3 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R65] 65.Timms RT et al. A glycine-specific N-degron pathway mediates the quality control of protein N-myristoylation. Science 365(2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R66] 66.Fischer ES et al. The molecular basis of CRL4DDB2/CSA ubiquitin ligase architecture, targeting, and activation. Cell 147, 1024–39 (2011). [DOI] [PubMed] [Google Scholar]

PERMALINK

CAND1 inhibits Cullin-2-RING ubiquitin ligases for enhanced substrate specificity

Kankan Wang

Stephanie Diaz

Lihong Li

Jeremy R Lohman

Xing Liu

Abstract

INTRODUCTION

RESULTS

CAND1 inhibits the degradation of CRL2 substrates

Fig. 1. CAND1 binds unneddylated CUL2 and inhibits the degradation of CRL2 target proteins.

Effects of neddyation and deneddylation on BRD2 degradation

Fig. 2. Eliminating neddylation, but not de-neddylation, inhibits ARV-771 induced degradation of BRD2.

CAND1 does not promote exchanges of CRL2 substrate receptors

Fig. 3. Dynamic assembly and disassembly of CRL2s in human cell lysate.

Binding kinetics of CRL2VHL and CUL2•CAND1

Fig. 4. Characterization of CUL2•VBC and CUL2•CAND1 interactions.

CAND1 strongly affects the koff of CRL2VHL

Fig. 5. Effects of CAND1 on the kon and koff of the CUL2-VBC interaction.

VBC alone marginally affects the kon or koff of CUL2•CAND1

Fig. 6. Substrate receptor module alone does not accelerate the dissociation of CUL2•CAND1.

CAND1 promotes CRL2 disassembly but not CRL2 assembly

CAND1 enhances CRL2 substrate specificity

Fig. 7. The accelerated disassembly of CRL2 by CAND1 has a stronger impact on the degradation of substrates that bind CRL2s with lower affinity.

DISCUSSION

METHODS

Cell lines

Generation of stable cell lines

Materials and plasmids

Antibodies

Protein expression and purification

Synthesis and purification of GGGGKAMCA peptide

Degradation of cellular proteins

Dissociation of CUL2•VBC in human cell lysate

Dissociation of CUL2•CAND1 in human cell lysate

Estimation of cellular VHL concentration

Dissociation of CUL2 from CAND1 expressed in different cells

Pulldown assay for CUL2•CAND1 dissociation

Biolayer interferometry assay

FRET-based assays

SILAC sample preparation

Mass spectrometry data collection

Mass spectrometry data analysis

Statistics and reproducibility

Extended Data

Extended Data Fig. 1. ARV-771 induced degradation of BRD2 in the presence of elevated levels of VHL.

Extended Data Fig. 2. Effects of CSN5i-3 on the degradation of CODD in the WT and DKO cells.

Extended Data Fig. 3. Assessing dynamic assembly and disassembly of CRL2s in human cell lysates.

Extended Data Fig. 4. Quantitative analyses of interactions between CUL2 and VBC, and between CUL2 and CAND1.

Extended Data Fig. 5. Interactions between CUL2 and VBC in the presence of CAND1.

Extended Data Fig. 6. CUL2•CAND1 dissociation and its effect on CUL2 neddylation.

Extended Data Fig. 7. ZBC hardly accelerates the disassembly of CUL2•CAND1.

Extended Data Fig. 8. PROTAC induced degradation of BRD2 and BRD3 in WT and DKO cells.

ACKNOWLEDGEMENTS

Footnotes

DATA AVAILABILITY STATEMENT

REFERENCES

METHODS-ONLY REFERENCES

Associated Data

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases

Binding kinetics of CRL2^VHL and CUL2•CAND1

CAND1 strongly affects the k_off of CRL2^VHL

Fig. 5. Effects of CAND1 on the k_on and k_off of the CUL2-VBC interaction.

VBC alone marginally affects the k_on or k_off of CUL2•CAND1

Synthesis and purification of GGGGK^AMCA peptide