Chemical profiling of DNA G-quadruplex-interacting proteins in live cells

Protein–nucleic acid interactions underpin key cellular processes, including transcription, DNA replication and DNA repair. Existing methods to study chromatin-associated protein complexes (for example, ChIP coupled to MS) often require high-quality antibodies and typically focus on one protein at a time, while enzyme-catalysed proximity labelling (BioID, APEX) can suffer from slow kinetics, toxicity, and large fusion tags. Photoactivatable small-molecule crosslinkers enable rapid, spatially controlled labelling in cells, but have mainly profiled direct small-molecule targets rather than broader interaction networks. DNA G-quadruplexes (G4s) are non-canonical four-stranded DNA structures that form at specific G-rich sequences and exist dynamically in human chromatin, especially at open chromatin and promoters of highly expressed genes. Although hundreds of thousands of potential G4-forming sequences exist in vitro, only a small fraction fold in chromatin in a cell-type-specific manner, suggesting that protein interactions regulate G4 formation and function. Prior identification of G4-binding proteins has largely relied on affinity pull-downs with synthetic G4 oligos from lysates, which do not capture the native chromatin context. The authors therefore set out to develop an in situ chemical strategy to map the interactome of endogenous DNA G4 structures in living cells with minimal perturbation.

ChIP–MS approaches have characterized chromatin complexes but depend on antibodies and single-target focus. Proximity labelling enzymes (BioID/TurboID/APEX) allow compartment- or protein-centric mapping but are constrained by labelling kinetics, toxicity, and tag size. Photoaffinity labelling with small molecules offers temporal control and lower background, and activity-/affinity-based profiling has mapped drug targets in cells, though not complex interaction networks. Foundational G4 studies (G4-seq and G4 ChIP-seq) established widespread potential G4 motifs and enrichment of folded G4s at regulatory chromatin, with cell-type specificity and promoter localization, implicating functional roles in transcription. Multiple proteins (helicases, transcription factors, epigenetic regulators) bind G4s in vitro, yet most discovery has used lysate-based pull-downs that lack chromatin context and may miss transient interactions. These gaps motivate a live-cell, structure-centric profiling strategy for G4 interactors.

The authors devised a co-binding-mediated protein profiling (CMPP) strategy using small-molecule G4 ligands functionalized with a photoreactive diazirine and an alkyne handle for click chemistry. Two probes based on the G4 ligand pyridostatin (PDS) were synthesized: photoPDS-1 (1) with a short two-carbon linker and photoPDS-2 (2) with a longer PEG linker (12 atoms), plus a control probe (3) lacking the G4-binding moiety. Biophysical characterization included FRET melting assays and fluorescence quench binding showing that 1 and 2 stabilize and bind G4 oligonucleotides (Myc, Kit1, Telo) selectively over dsDNA, with apparent Kd values comparable to PDS; control 3 showed no G4 binding. Proof of concept in vitro used the G4-specific antibody BG4 incubated with folded G4 Myc versus non-G4 controls (ss mutMyc, ds Myc). After adding probes and 365 nm irradiation, TAMRA-azide click conjugation, and SDS-PAGE with in-gel fluorescence, they observed dose-dependent labelling of BG4 only in the presence of G4 and only for probes 1 and 2, demonstrating co-binding-dependent photocrosslinking. In cells, HEK293T cells were treated with probes (typically 20 μM, 1 h), irradiated at 365 nm for 10 min on ice, and nuclear extracts prepared. For gel-based profiling, TAMRA-azide click and SDS-PAGE showed specific probe-dependent labelling and good nuclear uptake without detectable cytotoxicity. For proteomics, probe-labelled proteins were clicked to biotin-azide, enriched on streptavidin beads, on-bead digested, and analysed by label-free LC-MS/MS. Proteins enriched over control probe 3 (fold change >2, FDR <0.05) in at least 2 of 4 biological replicates were considered candidates. Probes 1 and 2 yielded 248 and 209 enriched proteins, respectively, with ~96% overlap (201 shared), indicating linker length was not critical. Functional annotation highlighted roles in transcription, RNA processing/splicing, and chromatin regulation. Validation of candidate binding used affinity enrichment from nuclear lysates with immobilized 3′-biotinylated G4 oligonucleotides representing different topologies (parallel: Myc, Kit1, Kit2; antiparallel: TBA; hybrid: BCL2), with ss and ds controls. Western blots confirmed sequence- and topology-selective G4 binding for 6/7 candidates (SMARCA4, UHRF1, RBM22, TTF2, DDX24, DDX1), while HMGB2 bound ss/dsDNA but not G4. Direct binding affinities were measured by ELISA with recombinant proteins: SMARCA4 bound G4 Kit1 (Kd 40.6 ± 5.1 nM); UHRF1 bound G4 Kit1 (Kd 1.2 ± 0.2 nM), tighter than its known substrates hemi-methylated dsDNA (8.5 ± 1.1 nM) and unmethylated dsDNA (21.2 ± 3.5 nM); DDX1 bound G4 Myc (5.1 ± 1.1 nM); DDX24 bound G4 Kit1 (58.2 ± 14.1 nM); RBM22 bound RNA NRAS G4 (52.1 ± 11.3 nM) and DNA G4s. To test binding in chromatin, SMARCA4 ChIP-seq was performed in K562 cells (three biological replicates; 28,265 high-confidence peaks). These were compared to previously mapped endogenous G4 ChIP-seq peaks (8,995 high-confidence). Overlap and signal profiles were computed, and genomic feature distributions were assessed. The SMARCA4 signal was enriched and centered at endogenous G4s but not at non-folded potential G4s, and 42% of shared peaks localized to promoters. Bioinformatics used cutadapt, BWA, Picard, MACS2, SAMtools, bedtools, deepTools, and MEME; proteomics statistics used limma/qPLEXanalyzer with imputation strategies detailed in Methods.

1. The CMPP probes (photoPDS-1 and -2) retained strong, selective G4 binding and stabilization compared to PDS, with negligible dsDNA stabilization; control probe 3 did not bind G4s. 2) In vitro photoproximity labelling showed dose-dependent crosslinking of the G4-binding antibody BG4 only when a G4 structure was present and only with G4-binding probes, confirming co-binding-mediated capture. 3) In HEK293T cells, CMPP identified hundreds of putative G4-interacting proteins: 248 (probe 1) and 209 (probe 2) enriched over control (FC>2, FDR<0.05); ~96% overlap (201/209) between probes indicated linker length was not critical. A subset overlapped known G4 interactors (probe 1: 19/79, 24%; probe 2: 11/79, 14%), and many novel candidates were discovered across functional classes, including transcription factors/regulators, RNA processing/splicing factors, helicases, and chromatin remodelers (e.g., SMARCA4, SMARCC1, UHRF1, TTF2, DDX1, DDX24, RBM22). 4) Affinity enrichment from nuclear lysates using diverse G4 oligonucleotides showed G4-specific binding for 6/7 tested proteins; HMGB2 bound ss/dsDNA but not G4s, suggesting indirect or adjacent DNA recognition. 5) Direct binding affinities (ELISA) demonstrated high-affinity, selective G4 binding: SMARCA4 to G4 Kit1 (Kd 40.6 ± 5.1 nM); UHRF1 to G4 Kit1 (Kd 1.2 ± 0.2 nM) versus hemi-methylated dsDNA (8.5 ± 1.1 nM) and unmethylated dsDNA (21.2 ± 3.5 nM); DDX1 to G4 Myc (5.1 ± 1.1 nM); DDX24 to G4 Kit1 (58.2 ± 14.1 nM); RBM22 to RNA NRAS G4 (52.1 ± 11.3 nM) and DNA G4s, with negligible binding to corresponding mutants or dsDNA. 6) SMARCA4 ChIP-seq in K562 cells identified 28,265 high-confidence binding sites; 84% of endogenous G4 peaks (7,565/8,995) overlapped SMARCA4 binding. SMARCA4 signal was strongly centered at endogenous G4s but not at non-folded potential G4 sites, indicating recognition of folded G4 structures rather than the underlying G-rich sequence. 7) Co-localization analyses showed the largest proportion of SMARCA4–G4 overlap at promoters (~42%), supporting a potential role in transcriptional regulation.

CMPP provides an in situ, structure-centric chemical strategy to map protein interactors of endogenous DNA G4 structures within native chromatin, addressing limitations of antibody dependence and the perturbations introduced by lysis-based pull-downs. The rapid, photoinduced proximity labelling mediated by small-molecule G4 ligands captures nearby co-binding proteins, potentially including transient interactors that might be lost during conventional workflows. Application to human cells yielded a broad and functionally diverse G4 interactome, encompassing previously known and numerous novel candidates. Orthogonal in vitro validations confirmed that several newly identified proteins bind G4s with high affinity and topology selectivity. Genomic profiling linked SMARCA4, a SWI/SNF chromatin remodeler, to endogenous G4 sites in vivo, with enrichment at gene promoters, suggesting a role for G4 structures in recruiting or modulating chromatin remodeling complexes and transcriptional programs. The approach should generalize across cell types and may be adapted to RNA G4s or other nucleic acid structural features using appropriate ligands. Together, these results advance understanding of G4 biology by directly connecting structural DNA elements with their cellular protein interactors in a live-cell context.

The study introduces CMPP, a live-cell, chemical proximity-labelling strategy leveraging G4-selective, photoactivatable ligands to map protein interactors of endogenous DNA G-quadruplex structures in native chromatin. CMPP identified hundreds of G4-associated proteins, confirmed several novel high-affinity G4 binders (e.g., SMARCA4, UHRF1, DDX1, DDX24, RBM22), and demonstrated that SMARCA4 extensively occupies endogenous G4 sites, especially at promoters. These findings establish CMPP as a robust approach to chart interactomes of nucleic acid structural motifs and implicate G4–protein interactions in transcriptional regulation via chromatin remodeling. Future work should include perturbation studies (knockdown/overexpression of candidate proteins coupled with G4 ChIP-seq), application across diverse cell states and lineages, development of RNA G4-specific probes to dissect RNA G4–protein networks, and extension to other nucleic acid structures.

Although treatment times and concentrations were limited, G4 ligands can perturb the endogenous G4 landscape, potentially stabilizing weak or transient G4s, altering topology, or inhibiting certain protein–G4 interactions at higher concentrations. Prolonged ligand exposure can induce DNA damage and recruit damage-associated proteins, though no enrichment of DNA damage proteins was observed under the conditions used. Diazirine photocrosslinkers can exhibit off-target/background binding; a non-G4-binding diazirine control (probe 3) was included to filter such background. Affinity-enrichment assays cannot distinguish direct from complex-mediated interactions, necessitating orthogonal validation as performed. Differences from an independent contemporaneous study likely reflect cell-type and technical differences (e.g., nuclear fractionation focus and background controls).