mRNA Display Identifies Potent, Paralog-Selective Peptidic Ligands for ARID1B (2024)

The ARID1A and ARID1Bsubunits are mutually exclusivecomponentsof the BAF variant of SWI/SNF chromatin remodeling complexes. Lossof function mutations in ARID1A are frequently observed in variouscancers, resulting in a dependency on the paralog ARID1B for cancercell proliferation. However, ARID1B has never been targeted directly,and the high degree of sequence similarity to ARID1A poses a challengefor the development of selective binders. In this study, we used mRNAdisplay to identify peptidic ligands that bind with nanomolar affinitiesto ARID1B and showed high selectivity over ARID1A. Using orthogonalbiochemical, biophysical, and chemical biology tools, we demonstratethat the peptides engage two different binding pockets, one of whichdirectly involves an ARID1B-exclusive cysteine that could allow covalenttargeting by small molecules. Our findings impart the first evidenceof the ligandability of ARID1B, provide valuable tools for drug discovery,and suggest opportunities for the development of selective moleculesto exploit the synthetic lethal relationship between ARID1A and ARID1Bin cancer.

Identificationof ARID1B Peptidic Binders via mRNA Display

We first expressedand purified the ARID1A and ARID1B EHD2 domainsin insect cells. We observed that ARID1A/B EHD2 show similar thermalstability and adopt a helical fold in solution, as measured by circulardichroism (CD) spectroscopy (Figure S2A,B) and in line with the published Cryo-EM data containing ARID1A.^5,6

With the aim of identifying tool compounds that could serveas the first proof of concept for the ligandability of ARID1B, weperformed an in vitro mRNA display selection formacrocyclic peptides using the RaPID technology⁷⁻⁹ (Figure ​Figure11A). The employed mRNA librariesencoded 10–14-membered macrocycles with a fixed C-terminalcysteine. Using the established flexizyme-mediated tRNA loading,⁸ we substituted the initiation methionine to N-chloroacetyl l-Phe for spontaneous cyclizationafter translation. Two different codon schemes were deployed againstARID1B containing a mixture of natural and unnatural amino acids (Figure S2C). The exchange of amino acids in thecodon scheme further increases the chemical and conformational spacesampled by the mRNA library.^10,11

We observed significant enrichmentin the selectionprocess againstARID1B compared with the negative control. Selected hits were synthesizedby solid-phase peptide synthesis (SPPS) and their ability to engageARID1B was evaluated by differential scanning fluorimetry (DSF). Weidentified seven peptides, three containing natural amino acids onlyand four containing between two and five non-natural amino acids,which elicited a thermal shift higher than 2 degrees (Figure ​Figure11B,C).

ARID1B Ligands DisplayHigh Affinity and Selectivity at TwoBinding Sites

Next, we performed affinity measurements bysurface plasmon resonance (SPR) of the identified peptides to ARID1Aand ARID1B. Single-cycle kinetic experiments confirmed the bindingof the seven peptides in the nanomolar range (Figure S3A). When selecting peptides 4 and 6 for K_D determination, we observedthat these ligands displayed very high affinity (peptide 4: 17 nM and peptide 6: 0.8 nM) (Figure ​Figure22A). SPR measurements against ARID1A revealedthat six out of seven peptides displayed significantly decreased signalon ARID1A vs ARID1B, whereas only peptide 7 retainedbinding to ARID1A with comparable affinities (Figure S3B). With these results, we prioritized peptides 4 and 6 as tools for further characterizationand developed two fluorescent probes by introducing a Cy5 fluorophoreon a C-terminal extension (peptides 4-cy5 and 6-cy5). Importantly, the fluorophore did not interfere with binding, asevidenced by similar affinities obtained in fluorescence polarizationmeasurements (Figure ​Figure22B). We next set up a TR-FRET competitive binding assay to performcompetition experiments with all identified peptides to bin the peptidesaccording to overlapping or distinct binding sites with fluorescentprobes 4-cy5 and 6-cy5. We observed thatpeptide 4 competed with 4-cy5 with an IC₅₀ of 38 nM, and peptide 6 competed with 6-cy5 with an IC₅₀ of 20 nM. In contrast, peptide 4 did not show competition with probe 6-cy5,and peptide 6 did not compete with probe 4-cy5, suggesting two distinct binding sites on ARID1B. We then performedcompetition experiments with all other identified peptides and observedthat they competed with either peptide 4-cy5 or 6-cy5 with nanomolar IC₅₀ values, suggesting thatthey have similar or overlapping binding sites with the respectiveprobe (Figure ​Figure22C).The existence of two nonoverlapping binding sites was further confirmedby native mass spectrometry (MS) analysis. In the apo form, ARID1Bpresents significant conformational heterogeneity as evidenced bymultiple charge state envelopes at less than 4000 mass-to-charge ratio(Figure ​Figure22D). The additionof peptide 4 or 6 resulted in stochiometricbinding and a significant reduction in conformational heterogeneityand, importantly, the addition of both peptides resulted in heterotrimericcomplex formation and highly reduced heterogeneity (Figure ​Figure22D).

In summary, we identified two distinct bindingsites on ARID1Bthat can be engaged by potent and ARID1B-selective peptidic ligands.

Mapping Recognition Sites of ARID1B Ligands

Next, wesought to map the binding sites of the identified peptides, focusingon peptide 6 due to its higher affinity and excellentselectivity over ARID1A. We performed hydrogen–deuterium exchangemass spectrometry (HDX-MS) using the recombinant ARID1B EHD2 domainand compared the apo control state to ARID1B EHD2 saturated with 6. The resulting differential heatmap highlights areas inthe protein with changes in deuterium uptake when bound to 6 (Figures ​Figures33A and S4A,B). We were intrigued by the putative bindingsite region suggested in HDX-MS data as this site is proximal to thepreviously reported BC-box,¹² which iscritical for ARID1A/B protein stability and overlaps with the ARID1-BRG1binding interface in the SWI/SNF complex. Encouraged by our previousstudy showing that perturbing the structural integrity of ARID1B leadsto loss of cancer cell proliferation when the SWI/SNF complex assemblyis affected,¹³ we sought to orthogonallyconfirm the binding site of peptide 6 through photoaffinitylabeling and peptide mapping.

To identify the best position to insert a photoreactivegroup,we performed an alanine scan on peptide 6 and subsequentlytested these variants in the TR-FRET competitive binding assay againsttheir parental peptide probe 6-cy5 (Figure S5A). We identified several amino acids where the mutationhad a strong influence on binding affinity. For example, mutationof Phe1 to alanine (peptide 10) almost completely abrogatedARID1B binding, thus indicating the phenylalanine is directly interactingwith ARID1B (Figure S5A). Based on theseresults, we synthesized photoprobe 23 replacing Phe1with para-azido-phenylalanine. The affinity of theprobe to ARID1B was confirmed by the TR-FRET competitive binding assay(Figure S5B). Next, we performed photo-cross-linkingexperiments by incubating recombinant ARID1B with 20 or 100 μMof photoprobe 23 and triggered cross-linking by irradiatingwith UV light (305 nm) followed by an online pepsinolysis and MS/MSanalysis. Strikingly, we found three overlapping peptides (residues2074–2084, 2074–2085, and 2077–2085) with a stronglydecreased intensity in both the 20 and 100 μM treatment conditionscompared with the DMSO control (Figure ​Figure33B,D). Although we could not identify these peptidesas being an adduct of photoprobe 23 (likely due to variouscofragmentation of probe-peptide adducts that are challenging to matchto spectral databases¹⁴), the strong andconsistent decrease in intensity compared with the DMSO control neverthelesssuggests that the region of 2074–2085 is in direct proximityto or directly involved in the binding site of photoprobe 23. This is further corroborated by the HDX-MS data, where an overlappingpeptide (residues 2073–2084) was identified with a significantlyreduced deuterium uptake upon binding of peptide 6 (Figure S4A).

Encouraged by the strong agreementbetween the HDX-MS and photo-cross-linkingdata pinpointing to a binding site proximal to C2074, we rationallydesigned six ARID1B point mutation constructs aiming at potentiallydisrupting binding with peptide 6. The mutants were designedto either change the shape/polarity of the binding site by growing(A2120Q, C2074Y, V2113N) or shrinking hydrophobic side chains (I2078S),reversing the charge of E2116 (E2116R), or a combination of hypothesesin the form of a triple mutant (I2078S–V2113N-A2120Q) (Figure ​Figure33E). The single mutantswere shown previously to not impair protein stability by deep mutationalscanning.¹³

We expressed and purifiedthese six mutants and measured bindingby DSF (Figure S6) and estimated affinitiesby fluorescence polarization (Figure ​Figure33F). Interestingly, while I2078 and E2116 seem largelydispensable for binding, mutations in A2120, V2113, and C2074 stronglydecrease the T_m of the ARID1B-6 complex, resulting in >40-fold decrease in binding affinity orevennonmeasurable affinity for the C2074Y mutant. Likewise, the bindingof the triple mutant I2078S–V2113N-A2120Q to peptide 6 was completely abolished. Together, these data suggest thatC2074, V2113, and A2120 are involved in the interaction with peptide 6.

Lastly, as the mapped binding site is overlappingwith the largeinterface formed between ARID1A/B and BRG1,^5,6 weassessed whether peptide 4 or 6 could perturbBAF complex composition. Given the intrinsic inability of the peptidesto permeate cell membrane, we performed glycerol sedimentation experimentsin ARID1A null HT-1376 nuclear lysates and assessed coelution of BRG1or SMARCB1 together with ARID1B in the presence of either peptide(Figure S7). In our experimental setup,we could not observe changes in BAF complex composition upon peptideaddition. We speculate that such an effect is likely due to extensivecooperative binding within the BAF complex, thus setting a high barfor a small cyclic peptide to disrupt the preformed, stable BAF complex.

ProteinExpression and Purification

DNA encoding ARID1B(1565–2236)-His6 was obtained from GeneArt and cloned intothe pFastBac vector. ARID1B mutants were generated using the QuikChangeSite-Directed Mutagenesis Kit (Agilent Technology). The final expressionconstructs were confirmed by Sanger sequencing. Recombinant baculoviruseswere then generated using the Bac-to-Bac Expression System (Invitrogen).After several rounds of viral amplifications, expression was conductedby infecting Sf21 at a density of 2 × 10⁶ cells permL with 3% of virus stock for 48 h at 27 °C. Cells were harvestedby centrifugation at 4000g for 20 min, and resultingpellets were directly frozen at −80 °C.

Cell pelletswere suspended in 50 mM HEPES at pH 8.0, 500 mM NaCl, 4 mM Imidazole,and 1 mM TCEP supplemented with cOmplete protease inhibitor (Roche)and Benzonase (Merck). Cells were lysed with 20 passages on a Potter-Elvehjemhom*ogenizer (Douncer) and insoluble cell debris was removed by centrifugationfor 45 min at 40,000g. Clarified cell lysate wasthen incubated with 5 mL of prewashed TALON beads (ClonTech) for 2h rotating at 4 °C. His-tagged protein was eluted from the beadswith an imidazole step (1 M imidazole in 50 mM HEPES pH 8.0, 500 mMNaCl, 1 mM TCEP) over 6 column volumes. The pool was diluted 1:6 with50 mM HEPES pH 7.5, 1 mM TCEP before being loaded onto a HiTrapQ HP5 mL column (GE Healthcare) mounted on a KTA Pure chromatography system(GE Healthcare). The protein was eluted with a linear gradient from0 to 1 M NaCl in 50 mM HEPES pH 7.5, 1 mM TCEP. The fractions containingARID1B protein were pooled, concentrated with Amicon Ultra-15 30Kcentrifugal filter unit (Merck), and further purified on a HiLoadSuperdex 75 16/600 pg size exclusion column (GE Healthcare) in 50mM HEPES pH 7.5, 300 mM NaCl, 1 mM TCEP. Finally, the protein wasconcentrated in an Amicon filter unit (Merck), flash-frozen with dryice, and stored at −80 °C. The purity and concentrationof the protein samples were determined by LC-MS.

InVitro mRNA Display

The coding regionof the peptide library was designed as follows: The initiator 5′initiator AUG codon, encoding for the N-terminal chloroacetylatedamino acid, was followed by 8–12 fully randomized positionscomprising trimer oligonucleotide mixtures. The trimer oligonucleotidepool includes one distinct codon for every amino acid, except forMet and Cys. The degenerate region was followed by a TGT encodingfor the C-terminal cysteine, enabling cyclization. At the 3′,a sequence encoding for the fixed spacer peptide sequence (GSGGSG)was followed by an amber stop codon (TAG). The in vitro translation system was reprogrammed through Flexizyme-mediated geneticcode reprogramming as previously described. Briefly, N-chloroacetyl l-Phe was used in place of the initiator methionine. Applyingthis reprogrammed translation system to the mRNA template affordsa 10–14 amino acid thioether macrocyclic peptide library witheach peptide containing a C-terminal Gly-Ser linker. For the screeningcampaign on ARID1B, two different codon tables were used, with thefirst one just containing all natural amino acids except for methionineand the second one containing 4 N-methylated amino acids replacingnatural ones (Figure S2C). Starting withan initial round containing <10¹³ unique cyclic peptides,ARID1B-specific binders were enriched using the EHD2 domain of ARID1B(aa 1565–2236) with a biotinylated C-terminal Avi-tag immobilizedonto streptavidin-conjugated magnetic beads. Nonspecific binders wereremoved using the streptavidin-conjugated magnetic beads only, andthe binding stringency was increased in the latter rounds throughlonger incubation times and/or wash steps. Two parallel selectioncycles were performed with either low (150 mM NaCl) or high salt (300mM NaCl) concentrations. Thereby, the samples were subjected to bufferexchange by small desalting columns after in vitro translation andreverse transcription. All selection rounds were submitted for NGSto obtain the enriched peptide sequences. Upon analysis, about 20different sequences were picked for chemical synthesis. The sevenpeptides described in the publication were identified at variablelevels of enrichment in one or two of the four selection setups withdifferent codon tables and NaCl concentrations.

Peptide Synthesis

Fluorescence Thermal Shift Assay

Proteins (2 μM)and peptides (5 μM) were diluted in 50 mM HEPES, 150 mM NaCl,0.25 mM TCEP, and 1% (v/v) DMSO, pH 7.5, containing 2× SYPROOrange dye (Thermo Fisher Scientific, Waltham, MA). The protein solutionswere added to 384-well, thin-walled Hard-Shell PCR microplates (4titudeLtd., Oakley, U.K.) that were covered by optically clear adhesiveseals to prevent evaporation. Measurements were carried out with aCFX Opus 384 Real-Time PCR System (BioRad, Hercules, CA). The temperaturewas increased from 20 to 95 °C at 1 °C/30 s, and the fluorescenceintensity was measured with the excitation and emission filters setto 465 and 590 nm, respectively. The data were analyzed with the CFXMaestro Software (BioRad, Hercules, CA).

Surface Plasma Resonance

The surface plasmon resonance(SPR) experiments were carried out using a Biacore 8K optical biosensorand Series S sensor Chip SA (Cytiva, Marlborough, MA). The chips werewashed three times with 1 M NaCl/50 mM NaOH. Proteins were injectedat a flow rate of 5 μL/min in SPR immobilization buffer (50mM HEPES, 150 mM NaCl, 0.05% (v/v) Tween 20, 0.25 mM TCEP, and 0.05%(w/v) BSA, pH 7.5) for 875 s. The experiments were performed at 278K with a flow rate of 50 μL/min in an SPR running buffer (SPRimmobilization buffer containing 1% (v/v) DMSO). The tested analyteswere diluted in an SPR running buffer. After baseline equilibrationwith a series of buffer blanks, a DMSO correction was performed from1 to 3%. Eight analyte concentrations were injected sequentially (160s), with a short dissociation between each injection. The last injectionwas followed by a 2000 s dissociation step. The data were fitted withthe Single-Cycle Kinetic module of Biacore 8K evaluation softwareusing a 1:1 binding model.

Fluorescence Polarization

Proteins(starting at 1 μM)and peptides (1 and 10 nM for 6-cy5 and 4-cy5, respectively) were diluted in 25 mM HEPES, 150 mM NaCl, 1 mM TCEP,0.05% (v/v) Tween 20, 0.05% (w/v) BSA, 1% (v/v) DMSO, pH 7.5 and incubatedfor 30 min at RT in 384 black well plate (Greiner BioOne, flat bottom,low volume). Fluorescence polarization was then measured on a PherasStardevice (BMG LABTECH, Ortenberg, Germany) with a setting time of 0.2s, 200 flashes focal height of 12.3 mm, and a target mP at 60. The K_D values were estimated by fitting the polarizationdata (mP) versus [ARID1B] by nonlinear fit regression (GraphPad Prism).

Time-Resolved Fluorescence Resonance Energy Transfer

BiotinylatedProtein (ARID1b(1565–2236)-Avi-His) and Tb-anti-HISAntibody (Invitrogen# PV5895) were diluted at twice the final concentrationin assay buffer (25 mM HEPES, 150 mM NaCl, 1 mM TCEP, 0.05% (w/w)Tween 20, 0.05% (w/v) BSA, pH 7.3) and preincubated for 30 min atRT in a white 1536 well plate in the presence of the samples dilutedin DMSO. After the addition of an equivalent volume of assay buffercontaining the appropriate tracer, the incubation was carried outfor 30 min. Time-resolved Fluorescence Resonance Energy Transfer (TR-FRET)was then measured on a PheraStar device (BMG LABTECH, Ortenberg, Germany)with an HTRF module (excitation 337 nm, emission 620/665 nm) with17 flashes, 100 μs delay, and 400 μs integration. Thefinal conditions in the assay buffer were 15 nM biotinylated protein,0.5 nM Tb-anti-HIS Antibody, 1% DMSO, and 30 or 10 nM tracer for 4-cy5 and 6-cy5, respectively,

Native MS

Proteins were dialyzed (Thermo ScientificSide-a-lyzer MINI dialysis devices, 20K MWCO) overnight into freshlyprepared 200 mM ammonium acetate at pH 6.8 at 4 °C with gentleshaking on an orbital shaker. Protein concentrations after bufferexchange were determined on a NanoDrop One. Protein was diluted toa final concentration of 5 μM. Samples were infused into a QExactive UHMR hybrid quadrupole Orbitrap mass spectrometer (ThermoFisher Scientific) equipped with a static nanospray ion source usingCT-1 nanospray tips (Humanix Cellomics tip, 1 μm for plant cells).Instrument scan parameters were as follows: scan range 1000–10,000mass-to-charge ratio (m/z), no fragmentation,resolution 6,250, 10 microscans, AGC target 3e6, and maximum injecttime 100. Instrument source parameters were as follows: spray voltage1.5 kV, capillary temperature 150 °C, and S-lens RF level 10.

HDX-MS

ARID1B stock solution was prepared in 10 mMTRIS, 250 mM NaCl at pH 7.8. HDX-MS sample solutions were preparedby mixing 50 μL of 42 μM ARID1B plus 1 μL of DMSOsolution (±2.4 mM peptide) and 49 μL of sample buffer.The solution was incubated for 30 min prior to the initiation of HDXexperiments. HDX labeling was performed in the stock buffer preparedwith D₂O pD 7.8 (pH_read 7.4). An exchange reactionwas initiated by diluting 4 μL of ARID1B sample solution with27 μL of deuterated buffer. The protein concentration duringlabeling was 1.5 μM, and that of the ligand was 1.7 μMin the reaction mixture that contained 93% deuterated water and lessthan 1% DMSO. The labeling reaction mixture was incubated at 18 °Cfor varying times (30–10,000 s). These partially deuteratedsamples were then subjected to HDX-MS analysis as follows. The labelingreaction was quenched at a 1:1 ratio with a solution of 2 M urea,1 M TCEP pH 1.7 at 2 °C. The quenched solution was then injectedonto an integrated fluidics system containing an HDx-3 PAL liquidhandling robot and a temperature-controlled chromatography chamberset at 1.5 °C (Trajan Scientific), a Dionex Ultimate 3000 UHPLCsystem, and a Q-Exactive HFX hybrid quadrupole Orbitrap mass spectrometer(Thermo Fisher Scientific). Mobile phase A: 0.1% formic acid in waterand mobile phase B: 0.1% formic acid in 95% acetonitrile in water.The protein is first loaded onto a mixed protease column (NovaBioAssays,FPXII:Pepsin. 1:1 dimension and PN) at 100 μL/min for 3 minat 5 °C. The resulting peptides are trapped and desalted on aC18 trapping column (Acclaim 5 μm, PepMap 300, 1 mm × 15mm, Fisher Scientific). Once desalting is complete, the trap columnis placed in line with a separation column (Hypersil GOLD 1.9 μmparticle size, 50 × 1 mm, Fisher Scientific) and eluted witha gradient of 10–35% B over 7 min. MS experiments were acquiredover a scan range of 350–1500 m/z.

Photoaffinity Labeling

• Mashtalir N.; et al. Modular Organizationand Assembly of SWI/SNF Family Chromatin RemodelingComplexes. Cell2018, 175, 1272–1288.e20. 10.1016/j.cell.2018.09.032. [PMC free article] [PubMed] [CrossRef] [Google Scholar]
• Kadoch C.; et al. Proteomic and bioinformaticanalysis of mammalian SWI/SNF complexesidentifies extensive roles in human malignancy. Nat. Genet.2013, 45, 592–601. 10.1038/ng.2628. [PMC free article] [PubMed] [CrossRef] [Google Scholar]
• Mittal P.; Roberts C. W. M.The SWI/SNF complexin cancer—biology, biomarkersand therapy. Nat. Rev. Clin. Oncol.2020, 17, 435. 10.1038/s41571-020-0357-3. [PMC free article] [PubMed] [CrossRef] [Google Scholar]
• Helming K. C.; et al. ARID1B is a specificvulnerability in ARID1A-mutant cancers. Nat.Med.2014, 20, 251–254. 10.1038/nm.3480. [PMC free article] [PubMed] [CrossRef] [Google Scholar]
• He S.; et al. Structure of nucleosome-boundhuman BAF complex. Science2020, 367, 875–881. 10.1126/science.aaz9761. [PubMed] [CrossRef] [Google Scholar]
• Mashtalir N.; et al. A Structural Model ofthe Endogenous Human BAF Complex Informs DiseaseMechanisms. Cell2020, 183, 802–817.e24. 10.1016/j.cell.2020.09.051. [PMC free article] [PubMed] [CrossRef] [Google Scholar]
• Ishizawa T.; Kawakami T.; Reid P. C.; Murakami H.TRAP display: A high-speedselection method for the generation of functional polypeptides. J. Am. Chem. Soc.2013, 135, 5433–5440. 10.1021/ja312579u. [PubMed] [CrossRef] [Google Scholar]
• Goto Y.; Katoh T.; Suga H.Flexizymes for genetic code reprogramming. Nat. Protoc.2011, 6, 779–790. 10.1038/nprot.2011.331. [PubMed] [CrossRef] [Google Scholar]
• RAPID DISPLAY METHOD IN TRANSLATIONAL SYNTHESIS OF PEPTIDE. WO2011049157, https://patentscope.wipo.int/search/en/detail.jsf?docId=WO2011049157.
• Huang Y.; Wiedmann M. M.; Suga H.RNA Display Methodsfor the Discoveryof Bioactive Macrocycles. Chem. Rev.2019, 119, 10360–10391. 10.1021/acs.chemrev.8b00430. [PubMed] [CrossRef] [Google Scholar]
• Josephson K.; Ricardo A.; Szostak J. W.mRNA display:from basic principlesto macrocycle drug discovery. Drug DiscoveryToday2014, 19, 388–399. 10.1016/j.drudis.2013.10.011. [PubMed] [CrossRef] [Google Scholar]
• Li X. S.; Trojer P.; Matsumura T.; Treisman J. E.; Tanese N.MammalianSWI/SNF-A Subunit BAF250/ARID1 Is an E3 Ubiquitin Ligase That TargetsHistone H2B. Mol. Cell. Biol.2010, 30, 1673–1688. 10.1128/MCB.00540-09. [PMC free article] [PubMed] [CrossRef] [Google Scholar]
• Mermet-Meillon F.; et al. Protein destabilizationunderlies pathogenic missense mutations inARID1B. Nat. Struct. Mol. Biol.2024, 1–5. 10.1038/s41594-024-01229-2. [PubMed] [CrossRef] [Google Scholar]
• Wozniak J. M.; et al. Enhanced mapping of small-molecule bindingsites in cells. Nat. Chem. Biol.2024, 1–12. 10.1038/s41589-023-01514-z. [PubMed] [CrossRef] [Google Scholar]
• Smith T. P.; et al. Identification and engineering of potent cyclicpeptides with selectiveor promiscuous binding through biochemical profiling and bioinformaticdata analysis. RSC Chem. Biol.2024, 5, 12–18. 10.1039/D3CB00168G. [PMC free article] [PubMed] [CrossRef] [Google Scholar]
• Low J. K. K.; et al. mRNA display revealsa class of high-affinity bromodomain-bindingmotifs that are not found in the human proteome. J. Biol. Chem.2023, 299, 105482 10.1016/j.jbc.2023.105482. [PMC free article] [PubMed] [CrossRef] [Google Scholar]
• Franck C.; et al. Discovery and characterizationof cyclic peptides selective for theC-terminal bromodomains of BET family proteins. Structure2023, 31, 912. 10.1016/j.str.2023.05.009. [PubMed] [CrossRef] [Google Scholar]
• Boatner L. M.; Palafox M. F.; Schweppe D. K.; Backus K. M.CysDB: a human cysteinedatabase based on experimental quantitative chemoproteomics. Cell Chem. Biol.2023, 30, 683–698.e3. 10.1016/j.chembiol.2023.04.004. [PMC free article] [PubMed] [CrossRef] [Google Scholar]
• Weiss A.; et al. FGF401, A First-In-ClassHighly Selective and Potent FGFR4 Inhibitorfor the Treatment of FGF19-Driven Hepatocellular Cancer. Mol. Cancer Ther.2019, 18, 2194–2206. 10.1158/1535-7163.MCT-18-1291. [PubMed] [CrossRef] [Google Scholar]
• Boike L.; Henning N. J.; Nomura D. K.Advances in covalentdrug discovery. Nat. Rev. Drug Discovery2022, 21, 881–898. 10.1038/s41573-022-00542-z. [PMC free article] [PubMed] [CrossRef] [Google Scholar]