Proximity labeling has become an essential approach for mapping spatial proteomes and macromolecular interactions within living cells, yet conventional protocols depend almost exclusively on fusing bulky enzymatic or photocatalytic proteins to the target molecule. These large genetic fusions, frequently exceeding twenty to thirty kilodaltons in mass, present substantial risks of altering native protein conformation, obscuring functional interaction epitopes, or mislocalizing delicate cellular targets. Addressing this technological bottleneck, researchers led by Wuyue Zhou, Shashank Nagaraja, Ciaran P. Seath, and colleagues have reported the development of FlAsH-ID, an alternative photoproximity labeling method published in Nature Chemical Biology that localizes catalytic labeling activity using a sequence of just six amino acids. By pairing the classic Tsien tetracysteine motif with the biarsenical dye FlAsH, the approach achieves efficient spatial tagging driven by localized singlet oxygen generation upon green-light illumination while preserving the baseline physiology and native regulatory functions of sensitive proteins in both cell lines and primary immune cells.

The methodological framework of FlAsH-ID couples the spatial precision of genetic targeting with the temporal sharpness of light-triggered catalysis. The team established that the synthetic FlAsH dye, when bound to the tetracysteine motif consisting of the hexapeptide sequence cysteine-cysteine-proline-glycine-cysteine-cysteine, functions as an efficient photosensitizer capable of activating chemical probes. To delineate the operative chemical mechanism, initial experiments were performed in human embryonic kidney HEK293T cell lysates irradiated under four-hundred-and-fifty-nanometer blue light or green light across a panel of candidate biotinylated capture probes, including diazirine, aniline, azide, and phenol derivatives. While diazirine yielded no detectable tagging, biotin-aniline showed outstanding reactivity, producing a fourteen-to-one signal-to-noise ratio over controls lacking the FlAsH catalyst and a fifty-nine-to-one ratio over dark controls. Mechanistic investigations using purified recombinant proteins—including carbonic anhydrase, bovine serum albumin, lysozyme, and ovalbumin—followed by tandem mass spectrometry identified widespread biotin-aniline adducts at histidine and tyrosine residues, demonstrating that localized singlet oxygen generation predominates alongside secondary single-electron transfer pathways, a finding corroborated by the complete loss of labeling upon addition of antioxidant scavengers such as Trolox and sodium ascorbate.

Evaluating the compatibility of biarsenical reagents within intact biological systems formed a crucial milestone in validating the technology for live-cell deployment. Although prolonged continuous exposure to micromolar concentrations of unbound FlAsH caused cytotoxicity over forty-eight hours with an observed half-maximal inhibitory concentration of 2.21 micromolar in HEK293T cells, a brief one-hour incubation followed by sequential ethanedithiol washing steps completely mitigated cellular harm, leaving viability unaffected after twenty-four hours. Comprehensive multi-parameter evaluations tracking markers of oxidative stress, cellular reactive oxygen species via CellROX assays, lipid peroxidation, and intracellular glutathione depletion via luminescent GSH-Glo measurements confirmed that localized FlAsH complexes caused no observable physiological damage over periods spanning up to forty-eight hours. Subsequent stability assays revealed that once the FlAsH fluorophore was coordinated to tetracysteine tags on chromatin-associated histone H3.1, nuclear fluorescence and biotinylation competence remained stable over twenty-four hours, demonstrating that the labeling machinery can remain in place to monitor dynamic cellular adaptations that unfold over prolonged experimental timeframes.

To rigorously quantify the effective spatial labeling radius of FlAsH-ID, the investigators engineered human lamin A constructs bearing the tetracysteine tag and applied two distinct super-resolution microscopy platforms. Using direct stochastic optical reconstruction microscopy coupled with a bioorthogonal trans-cyclooctene and tetrazine click chemistry workflow, the authors avoided the volumetric distortion of standard antibody sandwiches, calculating a pair correlation decay length that yielded a precise labeling radius of 45.4 plus or minus 1.8 nanometers. In parallel, expansion microscopy achieving an isotropic physical enlargement factor of approximately 5.67-fold allowed individual labeling puncta at the nuclear envelope to be resolved, yielding a quantified radius of 71.0 plus or minus 3.0 nanometers based on half-maximum peak width measurements. The observed difference between modalities reflected the anticipated linkage error introduced by streptavidin fluorophore conjugates during expanded sample processing, while comparisons against traditional miniSOG protein fusions demonstrated that miniSOG produced an effective labeling radius roughly ten nanometers wider alongside aberrant nuclear localization patterns.

The general versatility of the platform was systematically benchmarked across diverse subcellular compartments by targeting the tetracysteine tag to the nucleus via histone H3.1, the nucleolus via targeting sequences from the NIK protein, the cytosol via HINT1, the mitochondrial matrix via pyruvate dehydrogenase kinase 1, and the plasma membrane via B7-1 and epidermal growth factor receptor. In quantitative tandem mass tag chemoproteomics workflows profiling histone H3.1 interactomes after five minutes of green-light exposure, the platform identified 1,460 significantly enriched proteins, capturing fifty-three percent of the known nuclear proteome and ninety percent of curated BioGRID interactors, with chromatin components dominating the top hits. Applying the method to mitochondrial matrix kinase PDK1 in HEK293T cells yielded 222 significantly enriched factors with a mitochondrial proteome specificity of ninety-one percent, matching or exceeding established methods such as APEX, TurboID, and BioID while recovering canonical partners PDK2 and PDK3. Labeling of transmembrane EGFR successfully retrieved vesicle-trafficking networks and essential endocytic AP2 complex subunits alongside cytoplasmic signal transducers such as SHC1 and SRC, while targeting cytoplasmic major vault protein enriched canonical vault constituents like PARP4 and immune regulator MYD88.

Beyond static compartmental mapping, the prolonged retention of coordinated FlAsH permitted high-resolution tracking of dynamic proteomic reorganizations in response to acute biological stimuli. In human U-2 OS osteosarcoma cells stably expressing tagged histone H3.1, the labeling catalyst was loaded and washed eighteen hours prior to stimulating the cells with polyinosinic:polycytidylic acid to simulate viral double-stranded RNA exposure. Subsequent green-light irradiation and quantitative mass spectrometry captured 229 proteins exhibiting significantly increased nuclear enrichment following immune activation, of which 190 were annotated RNA-binding proteins cataloged in the RBP2GO database. Among the prominently enriched factors were validated antiviral response proteins such as TIAL1, CAPRIN1, FUBP3, BANF1, and PABPC1, as well as core exon junction complex components including CASC3, EIF4A3, and RBM8A. Orthogonal immunofluorescence assays confirmed that double-stranded RNA stimulation triggered a marked translocation of CASC3 from diffuse cytoplasmic distribution into distinct nuclear speckles, illustrating the utility of the method for capturing stimulus-dependent protein trafficking in native cellular environments.

The biological advantages of replacing bulky enzymatic tags with a minimal hexapeptide were most starkly demonstrated through functional investigations of RUNX3, a forty-four-kilodalton lineage-defining transcription factor governing CD8+ T lymphocyte differentiation and sensory neuron development. Because RUNX3 depends on its intact carboxy-terminal VWRPY motif to recruit transducin-like enhancer of split corepressors, any carboxy-terminal fusion disrupts binding, necessitating amino-terminal modifications. When the authors assembled an extensive comparative panel of amino-terminally tagged constructs spanning from the six-amino-acid tetracysteine tag up to the thirty-five-kilodalton TurboID enzyme, primary murine CD8+ T cells transduced with TurboID, miniTurbo, or ultraID displayed severely distorted differentiation profiles under varying interleukin-2 culture regimes. Flow cytometry revealed that while wild-type RUNX3 and tetracysteine-tagged RUNX3 faithfully upregulated CD103 and PD1 while suppressing CD62L and SLAMF6, enzyme-tagged variants induced aberrant hyper-elevation of CD69 and failed to induce PD1. Remarkably, even doubling tag length from six residues to a twelve-residue variant pushed cellular phenotypes toward an altered regulatory state, confirming that physical tag dimensions directly dictate functional fidelity.

Genome-wide bulk RNA sequencing of purified primary CD8+ T cells substantiated these phenotypic divergences at the transcriptional level. Unsupervised clustering and principal component analysis segregated wild-type RUNX3, the tetracysteine construct, and a small LOV-domain fusion into a coherent functional group, whereas large biotin ligase fusions formed an aberrant cluster characterized by neomorphic expression changes. Cells transduced with the six-amino-acid tagged construct exhibited differential expression of only ten genes relative to wild-type RUNX3 controls, whereas TurboID fusions failed to activate critical canonical cytotoxic effector genes, including perforin-1, granzyme B, and TIM-3, while inappropriately upregulating non-target transcripts. Structural modeling across fifty-six human proteins using AlphaFold3 corroborated these experimental observations, revealing that appending larger enzymatic modules systematically induced greater root-mean-square deviations and structural perturbations across core protein folds compared to compact tetracysteine insertions.

The physiological integrity of tagged RUNX3 constructs was subsequently tested in vivo using adoptive cell transfer into immunocompetent recipient mice infected with the Armstrong strain of lymphocytic choriomeningitis virus. Donor P14 CD8+ T cells transduced with wild-type RUNX3, tetracysteine-tagged RUNX3, TurboID fusions, or empty vector controls expanded vigorously through day eight post-infection, ruling out direct acute cytotoxicity from the fusion constructs. However, by day thirty post-infection, cells harboring TurboID fusions exhibited a more than ten-fold reduction in peripheral blood persistence compared to wild-type or tetracysteine-tagged populations and displayed defective formation of long-lived CD62L-high, CD127-high central memory T cells in the spleen. Instead, TurboID-expressing cells preferentially accumulated as short-lived terminal effector populations mirroring a Runx3-knockout defect, whereas the six-amino-acid tetracysteine tag preserved memory precursor generation, durable memory cell differentiation, and the rescue of Runx3-deficient donor T cell survival in living host animals.

Deploying FlAsH-ID within primary activated CD8+ T cells elucidated previously uncharacterized chromatin-regulatory partnerships that had remained obscured in conventional immunoprecipitation experiments due to the high insolubility of chromatin-bound RUNX3. Mass spectrometry analysis of labeled primary T cells identified 110 significantly enriched proximity partners, including core binding partner CBFB alongside prominent subunits of the SWI/SNF BAF chromatin-remodeling complex, specifically ARID1A, ARID1B, SMARCE1, BICRAL, and DPF2. Integrating these proteomic findings with genomic CUT&RUN, ChIP-seq, and ATAC-seq datasets revealed that over half of the genomic loci newly bound by RUNX3 upon T cell receptor activation simultaneously acquired ARID1A occupancy, with RUNX DNA motifs representing the most highly enriched sequence elements at co-occupied sites. ATAC-seq profiling of Runx3-deficient and Arid1a-deficient T cells demonstrated that genetic loss of either partner severely impaired chromatin accessibility at these regulatory elements, including crucial differentiation loci such as the interleukin-2 receptor alpha chain gene. Together, these findings establish that RUNX3 physically coordinates with BAF complexes to govern chromatin opening during immune activation, illustrating how minimal-tag photoproximity labeling can map native molecular interactomes of delicate transcriptional regulators without perturbing living cellular systems.

Subject of Research: Photocatalytic proximity labeling in live cells using a short tetracysteine peptide tag

Article Title: FlAsH-ID enables live-cell photoproximity labeling with a short peptide tag

Article References: Zhou, W., Nagaraja, S., Kim, J., Venables, T., Jones, M. T., Lee, S., Kang, J., Hacisuleyman, E., Pipkin, M. E., & Seath, C. P. (2026). FlAsH-ID enables live-cell photoproximity labeling with a short peptide tag. Nature Chemical Biology. https://doi.org/10.1038/s41589-026-02366-z

Image Credits: AI Generated

DOI: 10.1038/s41589-026-02366-z

Keywords: FlAsH-ID, proximity labeling, photocatalysis, tetracysteine tag, RUNX3, CD8+ T cells, singlet oxygen, proteomics, interactome, HEK293T cells, U-2 OS cells, mass spectrometry, super-resolution microscopy, dSTORM, expansion microscopy, chromatin remodeling, ARID1A, BAF complex