Cell culture

HEK293, HEK293T and HCT116 cells, originally sourced from the American Type Culture Collection, were provided by the Medical Research Council (MRC) Protein Phosphorylation and Ubiquitylation Unit (PPU) reagents facility at the University of Dundee. KBM7 iCas9 cells were a gift from J. Zuber (IMP Vienna). HEK293T, HCT-116 and Lenti-X 293T lentiviral packaging cells (Clontech) and Flp-In T-REx-293 (Thermo Fisher Scientific) cells were maintained in high-glucose DMEM (Gibco/Sigma-Aldrich). KBM7 cells were maintained in Iscove’s modified Dulbecco’s medium (IMDM; Sigma-Aldrich). Both DMEM and IMDM were supplemented with 10% FBS and 1% (v/v) penicillin–streptomycin (all supplied by Thermo Fisher Scientific/Sigma-Aldrich). Cell lines were grown in a humidified incubator at 37 °C and 5% CO2, routinely tested for Mycoplasma contamination and authenticated by short tandem repeat profiling.

Plasmids and oligonucleotides

Generation of the human CRL-focused sgRNA library used for SMARCA2 stability reporter screens, lentiviral sgRNA expression vector used for DCAF16 gene KO and viral vectors used for the engineering of inducible Cas9 cell lines was previously described26,35. The fluorescent protein stability reporter for SMARCA2 was generated by subcloning the BD (residues 1360–1534) of SMARCA2 from SMARCA2 pDONR223 (a gift from M. Taipale) and inserting it into a pRRL lentiviral vector, fused to a 3×V5 tag and mTagBFP at the C terminus and coupled to mCherry with a P2A self-cleaving peptide for normalization. The same SMARCA2 BD sequence was used to create a fusion construct with a miniTurboID enzyme48 at the N terminus, plus a flexible linker and a HiBiT tag. This construct was cloned into the pcDNA5/FRT/TO vector (V652020, Thermo Fisher), generously provided by the R. Hay lab, and verified by sequencing to ensure proper insertion and orientation.

The fluorescent protein stability reporter for BRD4 tandem BDs has been previously described35. To generate KO cell lines, DCAF16 and FBXO22 sgRNA was cloned into a pSpCas9(BB)-2A-EGFP vector from F. Zhang (PX458; Addgene, 48138). For reconstitution experiments, WT and cysteine mutants of FBXO22 fused to a 2×HA tag at the N terminus in a pLEX lentiviral vector were previously described26. Exogenous expression of SPIN4 was achieved using the same N-terminal 2×HA construct. Exogenous expression of DCAF16 was achieved from a pRRL lentiviral vector with a 3×Flag tag at the N terminus, as previously described35. Expression clones of DCAF16 mutants were generated from the WT 3×Flag tag vector using the Q5 site-directed mutagenesis (SDM) kit (New England Biolabs, E0552), according to the manufacturer’s instructions. For SDM studies, amino acids 2–216 of DCAF16 were selected for site saturation library design and cloning into a pRRL-SFFV-EGFP-cpHalo-(GSG)x9-DCAF16 backbone by GenScript. Quality control of the library by next-generation sequencing (NGS) was also performed by GenScript and resulted in a library coverage of 99.80% and 534×. For immunoprecipitation experiments, expression plasmids pRK5-HA-DCAF16 and pCMV5 HA-FBXO22 were kindly provided by X. Zhang (Northwestern University) or acquired from the MRC PPU reagents facility at the University of Dundee (DU42092). LgBiT and HaloTag–ubiquitin plasmids were purchased from Promega (N2681 and N2721 respectively). Flp-recombinase expression vector (pOG44) was kindly provided by the R. Hay lab. All plasmids and sgRNAs used in this study are shown in Supplementary Tables 5 and 6. The CRL-focused sgRNA library used for FACS-based CRISPR–Cas9 screens is shown in Supplementary Table 2.

Lentivirus production and transduction

Semiconfluent Lenti-X cells in 10-cm dishes were cotransfected with 1 µg of the envelope plasmid pMD2.G (Addgene, 12259), 2 µg of packaging plasmid psPAX2 (Addgene, 12260) and 4 µg of the lentiviral plasmid using polyethylenimine (PEI MAX, molecular weight: 40,000; Polysciences). Then, 3 days after transfection, supernatant containing virus was collected and filtered through a 0.45-mm filter. Target cells were infected in the presence of 8 μg ml−1 polybrene (szabo scandic, SACSC-134220).

Generation of monoclonal HEK293T KO cell lines

Semiconfluent HEK293T cells in six-well plates were transfected with 2.5 µg of a pSpCas9(BB)-2A-EGFP plasmid containing either DCAF16 or FBXO22 sgRNA (Supplementary Table 6) using Lipofectamine 2000 (Thermo Fisher Scientific, 11668027). Then, 2 days after transfection, single cells expressing EGFP were sorted into 96-well plates and allowed to grow for 2 weeks. Monoclonal cell colonies were verified for FBXO22 or DCAF16 KO by western blotting and preventing BRD4 degradation in the presence of IBG3 (ref. 35). As described with single-KO cell lines, double-KO clonal cells were generated starting from a clonal population of DCAF16-KO cells transfected with an FBXO22 sgRNA plasmid.

Generation of SMARCA HiBiT cell lines and KO clones

SMARCA2 and SMARCA4–HiBiT CRISPR–Cas9 knock-in cell lines were generated by ribonucleoprotein (RNP) transfection according to the manufacturer’s instructions for single-stranded DNA oligonucleotides (ssODNs; Integrated DNA Technologies, IDT). In brief, ssODNs served as donor templates, while recombinant SpCas9 Nuclease (IDT) was complexed with target-specific sgRNAs (Supplementary Table 6). For RNP transfection, 1 × 106 HEK293T cells were resuspended in MaxCyte buffer (cytiva) and mixed with SMARCA2/4-specific HiBiT ssODNs and RNPs. The mixture was filled into an electroporation cuvette to a final volume of 25 µl and electroporated with a MaxCyte ExPERT ATx. Following electroporation, cells were immediately transferred to a six-well plate as droplets and allowed to recover at 37 °C and 5% CO2 for 20 min. Following recovery, 2 ml of prewarmed DMEM (10% FBS and 1% penicillin–streptomycin) supplemented with Alt-R HDR Enhancer V2 (IDT) was added to the cells. The following day, the medium was exchanged and cells were allowed to expand. Then, 3 days after electroporation, the edited cells pools were analyzed with the HiBiT lytic assay (Promega, N3030) to detect luminescence and the HiBiT blotting system (Promega, N2410) to detect proteins tagged with HiBiT at the correct weight through western blotting. To obtain monoclonal populations, single cells were sorted into 96-well plates and allowed to grow for 2 weeks. Clones were expanded and validated as homozygous SMARCA2/SMARCA4–HiBiT clones by genotyping and by performing a lytic HiBiT degradation assay with the SMARCA-specific PROTAC, ACBI1 (ref. 49).

SMARCA4–HiBit HEK293T single-KO cell lines were generated by RNP electroporation using target-specific sgRNA (IDT; Supplementary Table 6) and spCas9 (CAS9PROT, Sigma-Aldrich). After forming the RNP complex through 20 min of incubation at room temperature, 4 × 105 HiBiT–SMARCA4 HEK293T cells were resuspended in Neon NxT Resuspension R Buffer (N10025, Thermo Fisher Scientific) along with the RNP complex and electroporated using a 10-µl Neon electroporation system cuvette tip (Thermo Fisher Scientific). Immediately following electroporation, the cells were placed in prewarmed DMEM supplemented with 10% FBS and 1% (v/v) penicillin–streptomycin. Then, 3 days after electroporation, individual cells were sorted into 96-well plates and allowed to grow for 2 weeks. To confirm the successful KO of the FBXO22 or DCAF16 genes, western blotting was used to verify the gene KO and to assess the prevention of BRD4 degradation in the presence of IBG3. Lastly, double-KO clonal cells were generated starting from a clonal population of FBXO22-KO cells that were subsequently electroporated with a DCAF16 sgRNA.

Western blotting

HEK293T and HCT116 cells were plated in six-well or 12-well plates depending on experimental setup at 0.5 × 106–0.6 × 106 cells per ml. Stock solutions of compounds were prepared in DMSO at a concentration of 10 mM and stored at −80 °C. In all experiments, the medium from cell seeding was exchanged with fresh DMEM containing working dilutions of experimental compounds. For the ubiquitin–proteasome inhibition assay, cells were cotreated for 18 h with 1 and 0.5 µM of TAK243, MLN4924 or carfilzomib.

For cell collection, cells were washed once with PBS before lysis on ice for 20 min with RIPA buffer (150 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS and 50 mM Tris-HCl pH 8) supplemented with benzonase (1:1,000; Sigma-Aldrich, 70746-3), HALT EDTA-free protease inhibitor cocktail (1:100; Thermo Fisher Scientific, 78437) and 20 mM DTT. After removal of the insoluble fraction by centrifugation at 15,000g at 4 °C for 15 min, supernatants were stored at −80 °C. Protein concentration was determined by bicinchoninic acid (BCA) assay (23225, Thermo Fisher Scientific). For HEK293, lysates were prepared with 4× Bolt LDS sample buffer (Thermo Fisher, B0008), heated at 95 °C for 5 min and then run on Bolt 4–12% Bis–Tris gels (Thermo Fisher Scientific) with Bolt MES SDS running buffer (Thermo Fisher Scientific). For HCT116, 4× NuPAGE LDS sample buffer was added to the cell lysates of (NP0007, Thermo Fisher Scientific) supplemented with 10% DTT and heated at 95 °C for 5 min. Samples (20–30 µg) were analyzed by protein electrophoresis with a NuPAGE MOPS SDS running buffer (Thermo Fisher Scientific). Proteins were transferred to nitrocellulose membranes (GE Healthcare, Amersham Protran supported 0.4 mm NC), blocked for 1 h with 5% milk in Tris-buffered saline with Tween-20 (TBS-T) at room temperature, before incubation with primary antibodies at 4 °C overnight. The following primary antibodies were used: HSP90 (1:1,000; Cell Signaling Technology, 4877), FBXO22 (1:200; Santa Cruz, sc-100736), SMARCA2 (1:1,000; Bethyl Laboratories, A301-015A), SMARCA4 (1:1,000; Bethyl Laboratories, A300-813A), HA epitope (3F10) (1:1,000; Merck, 11867423001), HA epitope (C29F4) (1:1,000; Cell Signaling Technology, 3724), HiBiT (1:1,000; Promega, N7200), IRDye 800CW Streptavidin (1:1,000; Licor, 926-32230), hFABTM rhodamine anti-tubulin (1:5,000; Bio-Rad, 12004165) and Flag (1:1,000; Sigma-Aldrich, F1804-200UG). Following overnight treatment, membranes were washed in TBS-T and incubated with horseradish peroxidase (HRP)-conjugated or fluorescence 800CW secondary antibodies for 1–2 h at room temperature. The following secondary antibodies were used: HRP anti-rabbit IgG (1:5,000; Cell Signaling Technology, 7074), HRP anti-mouse IgG (1:5,000; Cell Signaling Technology, 7076), IRDye 800CW anti-rabbit (1:5,000; Licor, 926-32211), IRDye 800CW anti-mouse (1:5,000; Licor, 926-32210), IRDye 800CW anti-rat (1:5,000; Licor, 926-32219) and HRP anti-mouse IgG (1:5,000; Cell Signaling Technology, 7076). Lastly, membranes were washed again with TBS-T and then imaged on ChemiDoc Touch imaging system (Bio-Rad) operated on Image Lab software (version 2.4.0.03). The biological conclusions of the western blots in this study were confirmed in multiple independent experiments. With the exception of Fig. 1c,e and Extended Data Figs. 1b, 5b and 6b,c, all western blot images are representative of 2–3 independent replicates. In addition, multiple orthogonal assays confirmed biological conclusions observed in western blots.

HiBiT degradation assays

SMARCA4–HiBiT cells were plated in 384-well or 96-well plates at a density of 3 × 103 or 3 × 104 cells per well, respectively. The following day, 5× stock solutions of compounds were dispensed either manually or with a Labcyte Echo550/555 liquid handler dispenser (Beckman Coulter Life Sciences). Cells were treated for 16, 18 or 24 h, as indicated in the respective figure legends before lysis using the HiBiT lytic assay buffer (Promega, N3030) according to the manufacturer’s instructions. Luminescence in each plate was measured using a PerkinElmer Victor X3 plate reader operated on PerkinElmer 2030 software (version 4.0). Treated wells were normalized to a DMSO-only control and analyzed using GraphPad Prism (version 10.0.3) through fitting of four-parameter nonlinear regression curves for extraction of DC50 values.

NanoBRET ubiquitination and kinetic degradation assay

For NanoBRET ubiquitination assays, 0.8 × 106 SMARCA2/4–HiBiT HEK293T cells were seeded in six-well plates. The following day, 1 µg of LgBiT (Promega, N2681) and 1 µg of HaloTag–Ubiquitin complementary DNA (cDNA; Promega, N2721) were transfected using FuGENE HD (Promega, E2311) at a 3:1 transfection reagent-to-plasmid ratio. After 8 h, cells were trypsinized and resuspended in phenol-red-free OptiMEM (Gibco) supplemented with 4% FBS and seeded in 96-well plates at a density of 0.2 × 106 cells per ml in the presence or absence of 0.1 mM HaloTag NanoBRET 618 ligand (Promega, G9801). Following overnight incubation, experimental compounds were added at the indicated concentrations. After 24 h of compound treatment, NanoBRET Nano-Glo substrate (Promega, N1571) was diluted in OptiMEM and added to cells at a 1× final concentration. The 96-well plates were analyzed using a PHERAstar (BMG Labtech) plate reader operated on PHERAstar software (firmware version 1.33) for NanoBRET ratio metric calculation (donor emission: 460 nm, acceptor emission: 618 nm). Data were processed by subtracting NanoBRET ligand-free controls before plotting the NanoBRET signal normalized to DMSO in GraphPad Prism (version 9.5.1).

Kinetic degradation assays were conducted using HiBiT-tagged SMARCA4 HEK293T cells that were transfected with exogenous LgBiT, following the same protocol as described for the NanoBRET ubiquitination assays. The cells were incubated with Endurazine substrate (1:100; N2570, Promega) for 2.5 h at 37 °C before the addition of 10× concentrations of experimental compounds. Luminescence measurements were taken every 5 min for a duration of 28 h using a GloMAX Discover microplate reader (Promega, software version 4.0.0, firmware version 4.92). The data were normalized to DMSO-only controls and T0 and then plotted as the luminescence signal over time using GraphPad Prism (version 9.5.1).

Generation of 293 Flp-In T-REx pools

A total of 1 × 106 Flp-In T-REx 293 cells (R78007, Thermo Fisher) were cotransfected using Lipofectamine 3000 with 3.6 µg of a Flp-recombinase expression vector (pOG44, Thermo Fisher Scientific) and 0.4 µg of a pcDNA5/FRT/TO plasmid containing a HiBiT–SMARCA2 BD with miniTurbo on the N terminus. Then, 24 h after transfection, polyclonal cell pools were washed, replaced with complete medium and, the following day, selected for 2 weeks with 200 µg ml−1 hygromycin B (Thermo Fisher Scientific, 10687010). Cells were expanded and then construct expression and biotinylation activity were induced with 1 µg ml−1 doxycycline (D9891, Merck) and biotin (B4501, Merck) to be validated by western blotting as described above.

BioID of SMARCA2 BD

Flp-In T-REx 293 stable cell line containing miniTurbo–SMARCA2 BD was divided into three treatment conditions, with each condition consisting of three 10-cm dishes at ~70% confluency. Expression of the fusion protein was induced by 1 µg ml−1 doxycycline treatment for a total of 20.5 h. Then, 12 h after doxycycline induction, cells were then pretreated with 10 µM MLN4924 neddylation inhibitor (5054770001, Merck) for 30 min, followed by addition of 10 µM 1, 2 or DMSO for 8 h. Biotin (100 µM final concentration) was added to cells undergoing treatment with compounds or DMSO for 2 h. At the endpoint, the culture medium was removed and cells were collected with PBS by scraping in 15-cm tubes pelleted by centrifugation at 600g for at least 5 min and washed for a second time in PBS. BioID was performed as described previously50,51. Cells were lysed with RIPA lysis and extraction buffer (25 mM Tris-HCl pH 7.6, 150 mM NaCl, 1% NP-40, 1% sodium deoxycholate and 0.1% SDS) (89900, Thermo Fisher Scientific) supplemented with complete EDTA-free protease inhibitor cocktail (11873580001, Roche) and 1:1,000 benzonase nuclease (70746-3, Sigma-Aldrich). Lysates were incubated at 4 °C for 15 min and then centrifuged at 13,000g for 15 min at 4 °C. Protein concentration was determined by BCA assay as described above. Next, 300 µg of protein lysate was incubated with 30 µl of precleared high-capacity streptavidin agarose beads (Pierce, 20359) and the mixture was incubated on a rotating wheel for 12 h at 4 °C. Beads were pelleted by centrifugation at 380g for 5 min and then transferred with 1 ml of lysis buffer to a fresh Eppendorf tube. Beads were washed five times with washing buffer (50 mM Tris-HCl and 0.5% NP-40) and eluted with 5% SDS.

Elutes were reduced using a final concentration of 10 mM Pierce DTT (A39255, Thermo Fisher Scientific) dissolved in triethylammonium bicarbonate buffer (TEABC) (T7408, Sigma Aldrich) at 60 °C for 30 min. Then, tubes were incubated to room temperature for 20 min in the dark and diluted to 20 mM iodoacetamide (A39271, Thermo Fisher Scientific) dissolved with TEABC. Afterward, samples were processed with an S-Trap mini spin column (PROTIFI, C02-mini) according to the manufacturer’s instructions. Lastly, tryptic digestion was performed by incubating the processed sample with 10 µg of Pierce trypsin protease MS-grade (90058, Thermo Fisher Scientific) dissolved in 100 µl of 100 mM TEABC overnight at 37 °C. Samples were then eluted from the column and transferred to a fresh Eppendorf tube. Columns were washed first with 0.15% formic acid and then with 50% acetonitrile in 0.15% formic acid and pooled with the eluate. Finally, samples were lyophilized and resuspended in 1% formic acid.

The dried peptides were reconstituted in 1% formic acid and analyzed on an Orbitrap Astral MS instrument connected to a Thermo Fisher Scientific Vanquish Neo ultrahigh-performance liquid chromatography (UHPLC) system. The peptides were enriched on a trap column and separated on an analytical column (Easy-Spray PepMap Neo C18, 2 μm, 75 μm × 150 mm) at 800 nl min−1. Chromatographic separation was performed using a gradient elution: 5% buffer A (0.1% formic acid) and 22.5% buffer B (90% acetonitrile and 0.1% formic acid) at 14 min and ramped to 35% at 21 min, followed by a return to 9%. The total run time was 22.6 min. Data acquisition was performed in data-independent acquisition (DIA) mode using the Astral analyzer. The MS data acquired in Orbitrap were operated with a fixed cycle time of 5 ms and with a full scan range of 380–980 m/z at a resolution of 240,000. The automatic gain control (AGC) was set to 500% and ion injection time is custom. Precursor ion selection width was kept at 4 Th and peptide fragmentation was achieved by higher-energy collisional dissociation (HCD, normalized collision energy 25%). For DIA mode, the scan range used was 150–2,000 m/z, AGC target was 500%, ion injection was custom and detector was Astral.

Quantitative proteomics

For unbiased identification of degrader target proteins, 1 × 106 HEK293 cells per condition were treated with DMSO (1:1,000), 1* (1 µM) or 1 (1 µM) for 12 h in biological triplicates. Cells were collected by centrifugation and washed two times in ice-cold PBS; cellular pellets were lysed in 100 µl of RIPA buffer supplemented with benzonase and cOmplete EDTA-free protease inhibitor cocktail. Cell debris was removed by centrifugation at 16,200g for 15 min at 4 °C. Supernatant was transferred to fresh tubes and protein concentration determined using the BCA protein assay kit. Next, 50 µg of cell lysate was then diluted to a final 5% SDS concentration and proteins were first reduced at 55 °C for 15 min in DTT (A39255, Thermo Fisher Scientific) dissolved in TEABC (T7408, Sigma-Aldrich) to a final concentration of 5 mM. Proteins were alkylated with to a final 20 mM iodoacetamide (A39271, Thermo Fisher Scientific) dissolved with TEABC in the dark for 10 min at room temperature. Afterward, samples were processed with an S-Trap micro spin column (PROTIFI, C02-micro) according to the manufacturer’s instructions. Lastly, tryptic digestion was performed by incubating the processed sample with 10 µg of Pierce trypsin protease MS-grade (90058, Thermo Fisher Scientific) dissolved in 100 µl of 100 mM TEABC overnight at 37 °C. Samples were then eluted from the column and transferred to a fresh Eppendorf tube. Columns were washed first with 0.15% formic acid and then with 50% acetonitrile in 0.15% formic acid and pooled with the eluate. Finally, samples were lyophilized and resuspended in 1% formic acid. The sample was analyzed on an Orbitrap Ascend Tribrid MS instrument coupled to a Thermo Fisher Scientific Vanquish Neo UHPLC instrument. The peptides were enriched on a trap column and resolved on an analytical column (Easy-Spray PepMap Neo C18, 2 μm, 75 μm × 150 mm) with 800 nl min−1. The gradient for separation was used as 2–7% B at 6 min, 7–18% at 89 min, 18–27% at 114 min and 27–35% at 134 min followed by column wash. The total run time used was 155 min. The data were acquired in data-independent acquisition mode in an orbitrap analyzer. The MS data acquired in Orbitrap were operated with standard AGC and auto ion injection time and with a full scan range of 300 to 1,350 m/z at a resolution of 60,000. For the DIA scan, the precursor mass range used was 400–1,000 m/z, precursors were isolated in quadrupole with an isolation window 10 Th and peptide fragmentation was achieved by HCD (normalized collision energy 30%). The resolution was set as 15,000.

MS analysis

Acquired raw files were processed using DIA-NN (version 1.8; https://github.com/vdemichev/DIaNN). Human UniProt was used as the protein sequence database. Two missed cleavages and a maximum of two variable modifications per peptide were allowed (acetylation of protein N termini and oxidation of methionine). Carbamidomethylation of cysteines was set as fixed modification. This data analysis was carried out using library-free analysis mode in DIA-NN with ‘deep learning-based spectra and RTs prediction’ enabled52. MBR was enabled. The search results were further processed in Perseus53. The data were loaded in Perseus in .txt format. Replicates were grouped on the basis of their annotation to 1, 2, 1* or DMSO. The data were log2-transformed and missing values were imputed from a normal distribution using the ‘processing → imputation → replace missing values from normal distribution’ function in Perseus. Imputation parameters included a width of 0.3 and a downshift of 1.8 s.d., simulating low-abundance values to replace missing entries. The imputation was applied in ‘whole matrix’ mode. Statistical analysis was performed using Perseus (version 2.0.11.0). Intensity values were log2-transformed before analysis. Differentially regulated proteins between conditions were identified using a two-sided permutation-based Student’s t-test with 250 randomizations. Results were visualized using volcano plots in GraphPad Prism (version 9.5.1) displaying log2 fold change versus −log10(P value).

SMARCA2 proximity labeling of DCAF16/FBXO22 with immunoprecipitation

Flp-In T-REx 293 miniTurbo–SMARCA2 BD-expressing cells were seeded in 10-cm dishes with 1.2 × 106 cells. The following day, cells were transfected with 5 μg of pRK5 HA–DCAF16 or pCMV5 HA–FBXO22. Then, 33 h after transfection, miniTurbo–SMARCA2 expression was induced with 1 µg ml−1 doxycycline for a total time of 15 h. Afterward, cells were pretreated for 30 min with 5 µM MLN4924, followed by the addition of 1 µM 1 or 2 and 170 µM biotin for 8 and 2 h, respectively. Cells were then washed in PBS and collected by centrifugation. The cell pellet was lysed in RIPA buffer supplemented with benzonase and cOmplete EDTA-free protease inhibitor cocktail. Next, 500 μg of cell lysate were incubated with 4 μg of HA antibody (51064-2-AP, Proteintech) or IgG control (sc-2026, Santa Cruz) overnight at 4 °C. Antibodies were then purified with 25 μl of protein A/G magnetic beads (88802, Thermo Fisher Scientific) for 1 h. Immunoprecipitated proteins were then washed five times in IP washing buffer (50 mM Tris-Hxl + 1% NP40 + 1% SDS) and eluted in 4× LDS sample buffer. Samples were analyzed by immunoblotting.

Flow-cytometric SMARCA2 reporter assays

KBM7 iCas9 cells were transduced with lentivirus expressing SFFV-SMARCA2 BD-mTagBFP-P2A-mCherry to generate stable reporter cell lines. Cell pools expressing both mTagBFP and mCherry were selected by sorting using a CytoFLEX SRT (Beckman Coulter) cell sorter operated on Beckman Coulter CytExpert SRT (version 1.1.0.10007) software. To quantify the influence of genetic perturbations on compound-induced reporter degradation, stable reporter cell lines were transduced with lentiviral sgRNA (pLenti-U6-sgRNA-IT-EF1αs-THY1.1-P2A-NeoR) and selected with neomycin. Cas9 expression was induced with doxycycline (0.4 µg ml−1) for 5 days, followed by 24 h of degrader treatment before flow cytometry analysis on an LSR Fortessa (BD Biosciences) operated on BD FACSDiva software (version 9.0). Data analysis was performed in FlowJo (version 10.10.0). The gating strategy is shown in Supplementary Fig. 3. BFP and mCherry mean fluorescence intensity values from Cas9-expressing cells were first normalized by background subtraction of the respective values from reporter negative cells. SMARCA2 abundance was calculated as the ratio of background subtracted BFP to mCherry mean fluorescence intensity and is displayed normalized to DMSO-treated cells.

FACS-based CRISPR–Cas9 SMARCA2 stability reporter screens

The SMARCA2 BD stability reporter FACS-based CRISPR–Cas9 screen, library preparation, NGS and data analysis were performed as previously described54. In brief, doxycycline-inducible Cas9 KBM7 cells stably expressing SMARCA2 BD-mTagBFP-P2A-mCherry were transduced in the presence of 8 μg ml−1 polybrene with a UPS-focused sgRNA library targeting 1,301 ubiquitin-associated human genes, with six sgRNAs per gene at a multiplicity of infection of 0.15 and over 1,000× library representation26 (Supplementary Table 2). Cells expressing sgRNAs were selected for 14 days with G418 (1 mg ml−1; Sigma-Aldrich, A1720) and then Cas9 expression was induced with doxycycline (0.4 μg ml−1; PanReac AppliChem, A2951). Next, 3 days after Cas9 induction, 50 × 106 cells were treated with DMSO, 0.1 μM 1 or 1 μM 2 for 24 h in two biological replicates.

After treatment, cells were washed once with 1× PBS and then incubated and stained with anti-Thy1.1–APC (1:400; BioLegend, 202526), Zombie NIR fixable viability dye (1:1,000; BioLegend, 423105) and human TruStain FcX Fc receptor-blocking solution (1:400; BioLegend, 422301) in FACS buffer (1× PBS, 5% FBS and 1 mM EDTA) for 10 min at 4 °C. Following two washes with FACS buffer, cells were fixed with BD CytoFix fixation buffer (BD Biosciences, 554655) for 45 min at 4 °C, protected from light. Fixed cells were then washed again with FACS buffer and stored overnight in FACS buffer at 4 °C. The following day, cells were strained trough a 35-µm nylon mesh and sorted using a BD FACSAria Fusion (BD Biosciences) operated on BD FACSDiva software (version 8.0.2), equipped with a 70-μm nozzle. Aggregates, dead (Zombie NIR-positive), Cas9-negative (GFP-negative) and sgRNA library-negative (Thy1.1–APC-negative) cells were excluded. The remaining cells were sorted on the basis of their SMARCA2 BD–BFP and mCherry levels into SMARCA2high (~5% of cells), SMARCA2mid (~30%) and SMARCA2low (~5%) fractions, ensuring a minimum library representation of 2,000× per replicate. Gating strategy is shown in Supplementary Fig. 3.

For library preparation, genomic DNA (gDNA) from the sorted cell fractions was isolated by cell lysis (10 mM Tris-HCl, 150 mM NaCl, 10 mM EDTA and 0.1% SDS), proteinase K treatment (New England Biolabs, P8107) and DNAse-free RNAse digestion (Thermo Fisher Scientific). DNA was then purified through two rounds of phenol extraction (Sigma-Aldrich, P4557) and isopropanol precipitation (Sigma-Aldrich, I9516). Isolated gDNA was subjected to a two-step PCR amplification of the sgRNA cassette using AmpliTaq gold polymerase (Thermo Fisher Scientific, 4311818), with sample-specific barcodes introduced during the first PCR and standard Illumina adaptors added in the second PCR. The resulting PCR products were purified using Mag-Bind TotalPure NGS beads (Omega Bio-tek, M1378-00) and the final Illumina libraries were pooled and sequenced on a NovaSeq 6000 platform (Illumina). Screen analysis was performed using the crispr-process-nf Nextflow pipeline (https://github.com/ZuberLab/crispr-process-nf/). Briefly, raw FASTQ files were trimmed with cutadapt (version 4.4) to remove random barcodes and spacer sequences. Demultiplexing was conducted on the basis of sample barcodes. Reads were aligned to the custom UPS sgRNA library using Bowtie2 (version 2.4.5) and guide abundance was quantified with featureCounts (version 2.0.1). Final read count tables were generated and subsequently analyzed using the crispr-mageck-nf workflow (https://github.com/ZuberLab/crispr-mageck-nf/) for downstream statistical analysis. Gene-level enrichment (fold changes and P values) was calculated in MAGeCK (version 0.5.9)55 by comparing sorted SMARCA2high or SMARCA2low populations against the SMARCA2mid reference population, using median-normalized read counts and replicate-level variance estimation.

DMS of DCAF16

To generate the lentiviral DCAF16 DMS library comprised of 4,300 DCAF16 single-point mutants fused to EGFP–cpHalo, Lenti-X HEK293T cells were seeded in 10-cm dishes and transfected at approximately 80% confluency with 4 μg of library plasmid, 2 μg of psPAX2 (Addgene, 12260) and 1 μg of pMD2.G (Addgene, 12259) using PEI (PolySciences). Viral supernatant was collected after 72 h and cleared of cellular debris by filtration through a 0.45-µm polyethersulfone filter.

Next, 50 million FBXO22 and DCAF16 double-KO HEK293T cells stably expressing SMARCA2 BD-mTagBFP-P2A-mCherry or BRD4-tandem BDs-mTagBFP-P2A-mCherry were transduced in the presence of 8 μg ml−1 polybrene with the DCAF16 DMS library virus at a multiplicity of infection of 0.16–0.19, yielding a calculated library representation of over 1,000 cells per variant. Library-transduced cells were selected with 1 μg ml−1 puromycin for 7 days. Approximately 20 million cells for each stability reporter were treated with DMSO, 1 μM 1, 1 μM 2, 1 μM 3, 1 μM MMH2 or 0.1 μM GNE-0011 for 24 h in three biological replicates. Following compound treatment, cells were trypsinized, harvested, washed with PBS and then fixed with BD CytoFix fixation buffer (BD Biosciences, 554655) for 40 min at 4 °C in the dark. Fixed cells were washed with PBS and stored in FACS buffer (1× PBS, 5% FBS and 1 mM EDTA) at 4 °C overnight.

The following day, cells were strained through a 35-μm nylon mesh and sorted on a BD FACSAria Fusion (BD Biosciences) operated on BD FACSDiva software (version 8.0.2) and equipped with a 70-μm nozzle. Aggregates, dead, reporter-negative (BFP-negative and mCherry-negative) and library-negative (EGFP-negative) cells were excluded from the sort. The remaining cells were sorted on the basis of their SMARCA2 BD–BFP or BRD4–tandem BDs–BFP and mCherry levels into low and high fractions (~5–15% of cells in each fraction), ensuring a minimum library representation of 1,500× per replicate. The gating strategy is shown in Supplementary Fig. 3. Three replicates of 20 million unsorted cells for each stability reporter were also harvested as controls, directly washed and frozen.

DNA libraries of sorted and unsorted cell pools for NGS were prepared as previously described54. In short, gDNA was extracted by cell lysis (10 mM Tris-HCl, 10 mM EDTA, 150 mM NaCl and 0.1% SDS), proteinase K treatment (New England Biolabs, P8107) and DNAse-free RNAse digestion (Thermo Fisher Scientific), followed by two rounds of phenol extraction and isopropanol precipitation. DCAF16 variant cDNAs were amplified by PCR from gDNA using Q5 high-fidelity DNA polymerase (New England Biolabs, M0491L) and the following primers: DCAF16_DMS_seq_F (TGGCACAGGAGGTTCAATG) and DCAF16_DMS_seq_R (TCATAATCCGCAGTTCCAGG). PCR reactions for each sample were pooled and purified using Mag-Bind TotalPure NGS beads (Omega Bio-tek, M1378-00). The library preparation from the amplified DNA was performed using the Tagment DNA TDE1 enzyme and IDT for Illumina unique dual indices (Illumina). Library concentrations were quantified with the Qubit 2.0 fluorometric quantitation system (Life Technologies) and the size distribution was assessed using the 2100 Bioanalyzer instrument (Agilent). For sequencing, samples were diluted and pooled into NGS libraries and sequenced on NovaSeq 6000 instrument (Illumina) following a 100-bp, paired-end recipe.

Raw sequencing reads were converted to FASTQ format with SAMtools (version 1.15.1) and BEDTools (version 2.30.0) using the bamtofastq function. Sequencing reads were trimmed using Trim Galore (version 0.6.6) using nextera and pair modes. Short reads were aligned to the DCAF16 cassette and SAM files were generated using the mem algorithm from the bwa software package (version 0.7.17). SAM files were converted to BAM using SAMtools and mutation calling was performed using the AnalyzeSaturationMutagenesis tool from GATK (version 4.1.8.1). Given the sequencing strategy, >98% of reads corresponded to WT sequences and were filtered out during this step. Next, relative frequencies of variants were calculated for each position and variants that were covered by <1 in 30,000 reads were excluded from further analysis. Read counts for each variant were then normalized to the total read count of each sample and log2 fold changes were calculated for sorted SMARCA2/BRD4high or SMARCA2/BRD4low fractions over unsorted pools. To correct for differential drug potency, each variant was then normalized to the maximum log2 fold change. For drug comparisons, log2 fold changes of SMARCA2/BRD4high or SMARCA2/BRD4low over unsorted pools of each drug were subtracted. Heat maps were generated using pheatmap (version 1.0.12) package in R (version 4.1.0).

Protein expression and purificationDCAF16 and mutants

The coding sequences for full-length DCAF16 with tobacco etch virus (TEV)-cleavable N-terminal His6 tags were cloned into a pFastBacDual vector under the control of the polh promoter. SDM PCR was used to create DCAF16 mutants (C58S, C58A, L59W, C173S and C173A). Coding sequences for full-length DDB1 or DDB1(ΔBPB) and full-length DDA1 were cloned into a pFastBacDual vector under the control of polh and p10 promoters, respectively. Bacmids was generated using the Bac-to-Bac baculovirus expression system (Thermo Fisher Scientific). Baculovirus was generated by adding bacmid (1 µg ml−1culture volume) mixed with 2 µg of PEI 25K (Polysciences) per µg of bacmid in 200 µl of PBS and incubated at room temperature for 30 min. The mixture was added to a suspension culture of Sf9 cells at 1 × 106 cells per ml in Sf-900 II SFM (Gibco) and incubated at 27 °C with shaking at 110 rpm. Viral supernatant (P0) was collected after 7 days. For expression in Trichoplusia ni High Five, cells were grown to densities between 1.5 × 106 and 2 × 106 cells per ml in Express Five SFM (Gibco) supplemented with 18 mM L-glutamine and infected with a total virus volume of 1% per 1 × 106 cells per ml, consisting of equal volumes of DCAF16 and DDB1 + DDA1 baculoviruses. For expression in SF9 cells, 3.5 × 106 cells per ml in SF-900 II SFM (Gibco) and infected with 1% of culture volume of each virus (DCAF16 and DDB1:DDA1). Cells were incubated at 27 °C in 2 L of Erlenmeyer flasks (~500 ml of culture per flask) with shaking at 110 rpm for 72 h. Cells were spun at 1,000g for 20 min and supernatant was discarded. Pellets were resuspended in lysis buffer (50 mM HEPES, 500 mM NaCl, 1 mM TCEP pH 7.5, Tween-20 to 1% (v/v), magnesium chloride to 2 mM, benzonase to 1 µg ml−1 and cOmplete EDTA-free protease inhibitor cocktail (Roche; two tablets per L of initial culture volume)); the resuspension was frozen and stored at −80 °C. Resuspended pellets were thawed and refrozen twice to help with cell lysis. Cell suspensions were sonicated, and lysates were centrifuged at 66,800g for 40 min. Clarified lysate was added to 2 ml of nickel agarose resin per L of culture on a roller at 4 °C for 1 h. The mixture of resin and lysate was centrifuged at 500g for 2 min to separate lysate from resin, taking the supernatant each time and washing the resin with wash buffer three times (50 mM HEPES, 500 mM NaCl, 1 mM TCEP and 20 mM imidazole, pH 7.5). Bound protein was eluted with elution buffer (50 mM HEPES, 500 mM NaCl, 1 mM TCEP and 500 mM imidazole, pH 7.5). Eluted protein was added to a dialysis bag (Snakeskin, 3.5-kDa molecular weight cutoff (MWCO); Thermo Fisher Scientific) and TEV protease was added to protein and incubated at 4 °C overnight. The sample was run over nickel agarose resin. Flowthrough and washes were collected and pooled. Protein was buffer-exchanged into ion exchange (IEX) buffer A (50 mM HEPES, 50 mM NaCl and 1 mM TCEP, pH 7.5) through 2 h of dialysis at room temperature with a change of dialysis buffer after 1 h. The sample was then loaded onto a HiTrap Q HP 5-ml column (Cytiva). The column was washed with IEX buffer A and bound protein was eluted with a 0–100% IEX buffer B (50 mM HEPES, 1 M NaCl and 1 mM TCEP, pH 7.5) gradient. Fractions containing protein were pooled, concentrated and run on a 16/600 Superdex 200 pg column or 10/300 Superdex 200 Increase (Cytiva) equilibrated in 20 mM HEPES, 150 mM NaCl and 1 mM TCEP, pH 7.5. Fractions containing the purified protein complex were pooled, concentrated, then aliquoted and flash-frozen in liquid nitrogen for storage at −80 °C.

Before intact protein MS experiments purified DCAF16–DDB1–DDA1 and mutants were dephosphorylated with a 1:40 ratio of Escherichia coli His–GST–lambda protein phosphatase (made in house) to DCAF16 protein at 4 °C overnight in 20 mM HEPES, 150 mM NaCl and 1 mM TCEP pH 7.5 buffer supplemented with 1 mM MnCl2. The next day, nickel resin (Abcam) was incubated with the mixture for 1 h, after which it was spun to sediment resin and recover the supernatant containing dephosphorylated DCAF16 and remove His–GST–lambda protein phosphatase that was immobilized on the resin.

SMARCA2

SMARCA2 BD (residues 1376–1590, Δ1403–1420) was PCR-amplified and subcloned into His12–SUMO (pRSF-DUET1) using quick ligase with 5′-BamHI and 3′-EcoRI. Plasmid was transformed into E. coli BL21 (DE3). Cultures were subjected to overnight expression at 18 °C, induced with 0.25 mM IPTG at an optical density of 600 nm (OD600) of ~0.8–1. Cells were collected by centrifugation and pellets were resuspended in lysis buffer (50 mM HEPES pH 7.5, 300 mM NaCl, 1 mM TCEP and 10% glycerol, supplemented with 2 mM magnesium chloride, benzonase and cOmplete EDTA-free protease inhibitor cocktail (Roche; one tablet per litre initial culture volume)); the resuspended pellet was frozen and stored at −20 °C. The resuspension was thawed and lysed at 30,000 psi using a CF1 cell disruptor (Constant Systems). The lysate was cleared by centrifugation at 50,400g for 30 min at 4 °C. The lysate loaded onto a 5-ml HisTrap HP column (Cytiva) equilibrated in lysis buffer and eluted with an imidazole gradient up to 100% elution buffer (50 mM HEPES, 500 mM NaCl, 0.5 mM TCEP and 500 mM imidazole, pH 7.5). Eluted sample was placed into a dialysis bag (Snakeskin, 3.5-kDa MWCO; Thermo) and ULP1 was added to sample to cleave the His12–SUMO tag and the bag was placed into 2 L of lysis buffer to dialyze overnight at 4 °C. SMARCA2 BD was run on a 5-ml HisTrap HP column equilibrated in lysis buffer. The flowthrough and wash containing SMARCA2 BD were pooled and concentrated in Amicon centrifugal filter units (3,000-kDa MWCO; Merck Millipore). The protein was loaded onto a HiLoad 16/600 Superdex 75 pg column (GE LifeSciences) equilibrated in 20 mM HEPES pH 7.5, 150 mM NaCl and 0.5 mM TCEP. Fractions containing pure SMARCA2 BD were confirmed by SDS–PAGE, then pooled, concentrated and aliquoted for storage at −80 °C until use. For GST-tagged SMARCA2 BD (residues 1376–1590, Δ1403–1420), SMARCA2 BD was subcloned into pDEST15 vectors (Invitrogen) and transformed into E. coli BL21 (DE3). Cultures were subjected to overnight expression at 18 °C, induced with 0.4 mM IPTG at an OD600 of ~0.8–1. Cells were collected by centrifugation and pellets were resuspended in lysis buffer (50 mM HEPES pH 7.5, 300 mM NaCl, 1 mM TCEP and 10% glycerol, supplemented with 2 mM magnesium chloride, benzonase and complete EDTA-free protease inhibitor cocktail (Roche; one tablet per L of initial culture volume)); the resuspended pellet was frozen and stored at −20 °C. The resuspension was thawed and lysed at 30,000 psi using a CF1 cell disruptor (Constant Systems). The lysate was cleared by centrifugation at 50,400g for 30 min at 4 °C. The lysate loaded onto a 20-ml of glutathione affinity resin (Abcam) equilibrated in lysis buffer and incubated on a roller for 3 h at room temperature. The resin was washed by multiple rounds of adding fresh lysis buffer. Elution buffer (50 mM HEPES pH 7.5, 300 mM NaCl, 1 mM TCEP, 10% glycerol and 25 mM L-glutathione) was added to GST–SMARCA2 BD-bound GST resin and incubated for 3 h on a roller at room temperature. The elution was collected and dialyzed into 50 mM HEPES pH 7.5, 50 mM NaCl, 1 mM TCEP and 10% glycerol, then subjected to a 5-ml Hitrap Q column and eluted on a 50 mM–1 mM NaCl gradient. GST–SMARCA2-containing fractions were loaded onto a HiLoad 16/600 Superdex 75 pg column (GE LifeSciences) equilibrated in 20 mM HEPES pH 7.5, 300 mM NaCl and 0.5 mM TCEP.

FBXO22–SKP1

His10–SUMO–SKP1–(GGS)×4–FBXO22 was Gibson-assembled into a pRSF-DUET1 plasmid. SDM PCR was used to create FBXO22 mutants (C228A and C326A). Protein expression was performed in BL-21 Rosetta (DE3) E. coli and overnight expression at 18 °C was induced with 0.25 mM IPTG at an OD600 of ~0.8–1. Cells were collected by centrifugation and pellets were resuspended in lysis buffer (25 mM HEPES pH 7.5, 250 mM NaCl, 2.5 mM TCEP and 20 mM imidazole, supplemented with 2 mM magnesium chloride, benzonase and cOmplete EDTA-free protease inhibitor cocktail (Roche, one tablet per L of initial culture volume)); the resuspended pellet was frozen and stored at −20 °C. The resuspension was thawed and lysed at 30,000 psi using a CF1 cell disruptor (Constant Systems). The lysate was cleared by centrifugation at 50,400g for 30 min at 4 °C. Clarified lysate was added to 2 ml of nickel agarose resin (Abcam) per L of culture on a roller at 4 °C for 1 h. The mixture of resin and lysate was centrifuged at 500g for 2 min to separate lysate from resin, taking the supernatant each time and washing the resin with wash buffer three times (50 mM HEPES, 500 mM NaCl, 1 mM TCEP and 20 mM imidazole, pH 7.5). An additional buffer wash was added with buffer supplemented with MgCl2 and ATP to remove heat-shock protein. Washed resin was resuspended in wash buffer and ULP1 was added to cleave the His10–SUMO tag and left to incubate overnight at 4 °C. The resin was placed into a Poly-Prep chromatography column (Bio-Rad) and the flowthrough and washes containing cleaved FBXO22-SKP1 were collected and pooled. Pooled protein was added to a dialysis bag (Snakeskin, 3.5-kDa MWCO; Thermo) and dialyzed into IEX buffer A (50 mM HEPES, 100 mM NaCl and 1 mM TCEP, pH 7.5) by 2 h of dialysis at room temperature with a change of dialysis buffer after 1 h. The sample was then loaded onto a HiTrap Q HP 5-ml column (Cytiva). The column was washed with IEX buffer A and bound protein was eluted with a 0–100% IEX buffer B (50 mM HEPES, 1 M NaCl and 1 mM TCEP, pH 7.5) gradient. Fractions containing protein were pooled, concentrated and run on a 10/300 Superdex 200 Increase (Cytiva) column equilibrated in 25 mM HEPES, 150 mM NaCl and 1 mM TCEP (pH 7.5). Fractions containing the purified protein complex were pooled, concentrated, then aliquoted and flash-frozen in liquid nitrogen for storage at −80 °C.

BRD4 BD2

BRD4 BD2 (residues 333-460) was expressed in E. coli BL21(DE3) and purified as described previously56. In brief, proteins were purified by nickel affinity chromatography and size-exclusion chromatography (SEC). His6 tag cleavage and reverse nickel affinity were performed before SEC for some applications; for others, the tag was left on. Purified proteins in 20 mM HEPES, 150 mM sodium chloride and 1 mM DTT (pH 7.5) were aliquoted, flash-frozen in liquid nitrogen and stored at −80 °C.

Intact protein MS

Before analysis, 10 µM recombinant human dephosphorylated DCAF16:DDB1:DDA1 (WT, C58S, L59W or C173S) was incubated with 20 µM DMSO or compounds and coincubated with 20 µM recombinant SMARCA2 BD for 18 h at room temperature. Buffer consisted of 25 mM HEPES pH 7.5, 150 mM NaCl and 1 mM TCEP. Samples were injected on a column (ZORBAX 300SB-C3), desalted for 1 min and then eluted to an UHPLC Agilent 1290 Infinity III high-throughput system (Agilent) using a gradient of 10% acetonitrile to 95% in 0.1% TFA and water. The fragmentor voltage was 135 V and the capillary voltage was 4,000 V. The MS instrument acquired full-scan mass spectra (m/z 600–3,000) on an InfinityLab Pro iQ Series Mass Detect. Mass spectra were deconvoluted using Agilent OpenLab CDS (version 2.8). Labeling efficiency for all intact mass analysis was calculated from relative abundance of the peaks.

Cryo-EM sample preparation

Before plunge freezing, 20 µM DDB1(ΔBPB)–DCAF16, 50 µM SMARCA2 BD and 60 µM 1 were incubated at ambient temperature for 1.5 h. The complex was diluted fivefold into 20 mM HEPES, 150 mM NaCl and 1 mM TCEP (pH 7.5). Before application on the grids, lauryl maltose neopentyl glycol (LMNG; Anatrace, NG310) was added to a final concentration of 0.01% (w/v). Then, 300-mesh copper Quantifoil R1.2/1.3 grids were glow-discharged for 60 s under vacuum at 30 mA using a Quorum SC7620 5 min before use. A total of 3.5 μl of the complex with 0.1% LMNG was applied to the carbon side only with final concentrations of 4 µM DDB1(ΔBPB)–DCAF16, 10 µM SMARCA2 BD and 12 µM 1. A Vitrobot Mark IV (FEI) (4 °C, 100% humidity, wait time = 10 s, blot force = 4, blot time = 4 s) was used for blotting with grade 595 filter paper (Agar Scientific, AG47000) before plunging into liquid ethane.

Cryo-EM data acquisition

All grids were clipped and screened at the University of Dundee EM Facility using a Glacios (Thermo Fisher Scientific) operating at 200 kV equipped with a Falcon4i direct electron detector and Selectris energy filter. The dataset for DDB1(ΔBPB)–DCAF16, SMARCA2 BD and 1 was collected at the UK National Electron Bio-Imaging Center (eBIC; Diamond Light Source) under BAG access BI-31827-25 using a Krios-II (m03) operating a Schottky X-FEG at 300 kV and Ametek-Gatan BioQuantum K3 imaging filter with slit width of 20 eV. Data were recorded using single-particle EPU version 3.8 (Thermo Fisher Scientific) with aberration-free image shift at a nominal magnification of ×165,000, calibrated pixel size of 0.513 Å, C2 aperture of 50 μm and objective aperture of 100 μm. All videos were collected using a Gatan K3 direct electron detector (5,760 × 4,092 pixels) in counting mode over 60 dose fractions, recorded as Tiff LZW non-gain-normalized, with a total dose of 60 e− per Å2, exposure time of 1.26 s and dose rate of 15.7 e− per pixel per s. In the first instance, single-particle EPU was used to collect 4,336 videos at 0° alpha tilt. For the high-tilt dataset, alpha tilt was set to 30° in single-particle EPU and 6,550 videos were collected under identical settings.

Single-particle cryo-EM data processing

Full videos and corresponding metadata were separately imported to cryoSPARC (version 4.5)57 for the 0° tilted and the 30° tilted datasets. Each dataset was subjected to patch motion correction and patch contrast transfer function (CTF) estimation and then manually curated with CTF estimation cut at 5 Å. For the 0° dataset, micrographs were denoised in cryoSPARC using 300 micrographs for training over 1,200 epochs. Unbiased particle picking was carried out with blob picker on denoised micrographs with a circular blob (minimum diameter of 80 Å and maximum diameter of 160 Å). The blob-picked particles were extracted with a box size of 512 pixels, Fourier-cropped to 128 pixels (pixel size = 2.05 Å, 4× binning) and subjected to two-dimensional (2D) classification to remove obvious junk. In the first round, the blob-picked particles were input into ab initio reconstruction with eight classes, followed by heterogeneous refinement. Many classes contained free DDB1(ΔBPB) or distorted volumes but one class did contain the complex of DDB1(ΔBPB)–DCAF16 and SMARCA2 BD. Particles for the complex were subjected to 2D classification to remove obvious junk particles and rebalanced by orientations; a subset of ~8,000 particles were taken for training in TOPAZ (version 0.2.5a)58. The trained TOPAZ models were used to extract particles from their respective micrographs, extracting 512 pixels at a pixel size of 2.05 Å. The particles were subjected to ab initio reconstruction with six classes. For the 30° dataset, following manual curation, a template picker using a good volume of the complex from a previous data collection was used and picks were inspected and subjected to 2D classification, where junk and free DDB1(ΔBPB) complexes were removed. Selected particles were taken for training in TOPAZ. The trained TOPAZ models were used to extract particles from their respective micrographs, extracting 512 pixels at a pixel size of 2.05 Å. The particles were subjected to ab initio reconstruction with four classes. Particles for the complex were subjected to 2D classification to remove junk and rebalanced by orientations; a subset of particles were taken for further training in TOPAZ. The particles were subjected to 2D classification, ab initio reconstruction (six classes) and nonuniform refinement of the best class. Particle sets were created from this and a further round of TOPAZ was run, followed by ab initio reconstruction into six classes. At this point, particles from ab initio reconstruction containing the complex were extracted at their full box size of 612 pixels (0.513 Å) in separate extraction jobs. The extracted particles from each dataset were combined for ab initio reconstruction in four classes, followed by heterogeneous refinement. Some junk classes with distorted volumes or incomplete complexes were present but a clear volume of the complex emerged and was taken for nonuniform refinement. From the best ab initio class, duplicate particles were removed. Particles were subjected to global CTF refinement followed by a nonuniform refinement, reconstruction and three-dimensional (3D) classification into five classes. The best class was reconstructed and subjected to further classification (four classes). The best volume from the 3D classification was imported to UCSF ChimeraX (version 1.8), Gaussian-blurred, segmented and resampled; an .mrc volume encompassing SMARCA2 and p